Handbook on the Physics and Chemistry of Rare Earths: High Temperature SuperConductors - II

Handbook on the Physics and Chemistry of Rare Earths, volume 31 High-Temperature Superconductors - II Elsevier, 2001 Edi...

Author: M.B. Maple | K.A. Gschneidner | L. Eyring

230 downloads 777 Views 14MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Handbook on the Physics and Chemistry of Rare Earths, volume 31 High-Temperature Superconductors - II Elsevier, 2001 Edited by: Karl A. Gschneidner, Jr., LeRoy Eyring and M. Brian Maple ISBN: 978-0-444-50719-8

Handbook on the Physics and Chemistry of Rare Earths Vol. 31 edited by K.A. Gschneidner, Jr., L. Eyring and M.B. Maple © 2001 Elsevier Science B.V. All rights" reserved

PREFACE K a r l A. G S C H N E I D N E R ,

Jr., L e R o y E Y R I N G , a n d M . B r i a n M A P L E

These elements perplex us in our rearches [sic], baffle us in our speculations, and haunt us in our very dreams. They stretch like an unknown sea before us mocking, mystifying, and murmuring strange revelations and possibilities.

Sir William Crookes (February 16, 1887)

This volume of the Handbook is the second of a two-volume set of reviews devoted to the rare-earth-based high-temperature oxide superconductors (commonly known as hiTc superconductors). The history of hiTc superconductors is nearly 15 years old. The initial breakthrough came late in 1986 when Bednorz and Miiller published their results which showed that (La,Ba)2CuO4_y had a superconducting transition of ~30K, which was about 7 K higher than any other known superconducting material. Within a year the upper temperature limit was raised to nearly 100 K with the discovery of an ~90 K superconducting transition in YBa2Cu307-x. The announcement of a superconductor with a transition temperature higher than the boiling point of liquid nitrogen set-off a frenzy of research on trying to find other oxide hiTc superconductors. Within a few months the maximum superconducting transition reached l l 0 K (Ba2Sr2CazCu3010), and then 122 K (TIBa2Ca3Cu4OI1). It took several years to push Tc another 11 K to 133 K with the discovery of superconductivity in HgBa2Ca2Cu3Os, which is still the record holder today. Because of the rapid development of our knowledge of these materials, a review on this topic several years ago would have been hopelessly out of date even before the paper would be sent to the publisher. About five years ago the field began to mature, and we felt it would be a good time to look into the possibility of publishing a series of review papers on rare-earth-containing hiTc superconductors. Because of the size of this project and the limited knowledge of the senior editors on this topic, we recruited Professor Brian Maple to assist us in undertaking this task. The result of our efforts have paid off as volumes 30 and 3l of the Handbook series. In this volume we have ten chapters to complement the eight which appeared in print as volume 30 at the end of 2000. The first chapter (195), by Emanuel Kaldis, is an extensive review of oxygen nonstoichiometry and lattice effects in YBa2Cu3Ox (often referred to as YBCO). As noted

vi

PREFACE

by the author, the oxygen content of YBCO which is significant for structural changes is also important for changes in other properties, i.e. more often than not there is a one-toone correlation of the superconducting properties with lattice defects. Kaldis points out that many of the difficulties in the preparation and characterization of the high-temperature superconductors is due to the extremely small coherence length in these materials which induces a large sensitivity to nano- and mesoscopic inhomogeneities. On the other hand, this property presents a great advantage for materials research since it can be used as a probe to investigate phase relationships at a very small length scale. The next chapter (196), by Harald W Weber, concentrates on flux pinning effects which result in high critical current densities even at high temperatures and in high magnetic felds, i.e. applications of YBCO-based materials near the boiling point of liquid N2. The influence of chemical substitutions is also reviewed, especially that of Nd. Although the main emphasis is on the characteristics of strong flux pinning, weak flux pinning is also reviewed. The nature of the defects which are responsible for flux pinning is reviewed, including those intentionally introduced by melt-texturing, or developed during the growth process, or artificially induced by suitable types of radiation. The magnetoresistance and Hall effect in both normal and superconducting states of the cuprate superconductors are reviewed in chapter 197 by Carmen C. Almasan and M. Brian Maple. These measurements address several issues: (1) in the normal state, the temperature dependence of the Hall coefficient and the cotangent of the Hall angle; and (2) in the superconducting state, the temperature dependence of the upper critical field and the irreversibility line, as well as the anomalous sign change of the Hall effect near the superconducting transition temperature. This information is valuable in understanding the electronic structure and various excitations in both states, and vortex pinning and dynamics in the superconducting phase. Chapters 198 and 199 are devoted to neutron scattering studies. In chapter 198 Thomas E. Mason describes neutron scattering investigations of spin fluctuations in the CuO2 planes. As noted by Mason, since many of the theories of high-temperature superconductivity include spin fluctuations as central feature of the mechanism for superconductivity it is important to obtain a detailed picture of the energy, momentum, temperature and doping dependence of the magnetic dynamics of the cuprates. But because of the small spin quantum number (½) and the large energy scale over which spin fluctuations exist (hundreds of meV), large single crystals are required (>1 cm3). As a result, only two systems, La2_x(Sr,Ba)xCuO4 and YBa2Cu3OT-x, have been studied in any detail by inelastic neutron scattering techniques. The second neutron scattering chapter (199), by Jeffrey W. Lynn and S. Skanthakumar, deals with the ordering of the lanthanide ions in various cuprates. In these cuprate materials, the lanthanide (R) ions are electronically isolated from the Cu-O layers where the Cooper pairs form and have low ordering temperatures (~1 K) which make them prototypical "magnetic superconductors". Furthermore, because of the layered crystal structure they are some of the best two-dimensional magnets known. However, in several of these magnetic superconductors, the magnetic ordering temperature is too high to be explained by dipolar interactions, and R-R exchange interactions must play a dominant role in the magnetism. The authors have also reviewed the behavior of the praseodymium ion in the cuprates, which differs from the other lanthanides. In the

PREFACE

vii

praseodymium cuprate, the 4f electrons are strongly hybridized with the conduction electrons which accounts for the high magnetic ordering temperature and the absence of superconductivity. Chapter 200, by Peter M. Allenspach and M. Brian Maple, reviews some aspects of the low-temperature heat capacity of the ceramic oxide superconductors. These measurements yield valuable information about the electronic, lattice, magnetic, crystalline electric field and hyperfine nature of the various rare-earth cuprate materials, and in that respect compliment other physical property studies, such as neutron diffraction, inelastic neutron scattering, and various spectroscopic measurements. The authors review the heat-capacity properties of the stoichiometric RBaCu307 compounds and oxygen-deficient materials, and show that there are significant differences. The heat capacities of other lanthanide cuprates, such as RBa2Cu408 and R2B4Cu7014+x, are also discussed. The next three chapters in this volume are concerned with various spectroscopies photoemission (chapter 201), infrared (chapter 202) and Raman (chapter 203). In chapter 201, Matthias Schabel and Zhi-xun Shen discuss the angle-resolved photoemission studies of untwinned YBa2Cu309_ x. This technique is able to directly probe both the energy and momentum scales of these highly correlated materials, and because of the two-dimensional nature of the cuprates, has unambiguously detailed their planar band structure. The infrared properties of the high-temperature oxide superconductors and knowledge obtained from such measurements are reviewed in chapter 202 by Dimitri N. Basov and Thomas Timusk. Infrared spectroscopic studies have yielded important information about the influence of doping the antiferromagnetic cuprate insulating phases with charge carriers; some novel features in both the charge dynamics in the normal state and the electrodynamics in the superconducting state; and the structure of vortices by magnetooptical measurement of thin films of YBCO. The authors also note Chat magnetooptical studies at high magnetic field may be useful in resolving the controversy about relaxation mechanisms in the cuprate superconductors. S. Lance Cooper's review (chapter 203) on the electronic and magnetic Raman scattering studies describes the contributions made by this technique to our understanding of the nature of the four distinct regimes of these copper-oxide-based materials as a function of oxygen doping. These are: the antiferromagnetic insulator at zero doping, the underdoped superconductor; the optimally doped superconductor; and the overdoped superconductor. Raman scattering is a powerful tool since it is sensitive to spin, electronic and phonon degrees of freedom, and thus it can provide information on energy, symmetry and lifetimes of various excitations which occur in these complex cuprate phase diagrams. It can also increase our knowledge of the nature of the superconducting pairing mechanism, how antiferromagnetic correlations evolve into the metallic phase, and the relationship between the pseudogap and the superconducting phases in the underdoped copper oxides. The tunneling spectra of the cuprate superconductors and the characterization of these materials by scanning tunneling microscopy are discussed in chapter 204 by Hiroharu Sugawara, Tetsuya Hasegawa and Koichi Kitazawa. Tunneling spectroscopy is a powerful tool for exploring the density of states near the Fermi level, and because its energy resolution (~<1meV) is higher than any other technique, it is crucial for determining

viii

PREFACE

the gap structures of superconductors. The authors note that the tunneling spectra have revealed some strange features: a finite differential conductance (and sometimes a peak) in the gap; and various energy dependences (including linear and parabolic) in the background spectra outside of the gap region. Cryogenic vacuum scanning tunneling microscopy (STM) allows one to demonstrate atomic resolution on the surface of materials in a chemical-free environment on the interface between the specimen and insulating layer. STM studies of the high-temperature superconducting cuprates have revealed many unique and unexpected features in these materials.

CONTENTS

Preface Contents

v ix

Contents of Volumes 1-30

xi

195. Emanuel Kaldis

Oxygen nonstoichiometry and lattice effects in YBaeCu3Ox. Phase transitions, structural distortions and phase separation 1 196. Harald W. Weber Flux pinning 187 197. C.C. Almasan and M.B. Maple

Magnetoresistance and Hall effect

251

198. T.E. Mason

Neutron scattering studies of spin fluctuations in high-temperature superconductors 281 199. J.W. Lynn and S. Skanthakumar

Neutron scattering studies of lanthanide magnetic ordering

315

200. Peter M. Allenspach and M. Brian Maple

Heat capacity

351

201. Matthias Schabel and Z.-X. Shen

Angle-resolved photoemission studies of untwinned yttrium barium copper oxide 391 202. D.N. Basov and T. Tinmsk

[nfrared properties of high-To superconductors: an experimental overview

437

203. S.L. Cooper

Electronic and magnetic Raman scattering studies of the high-Tc cuprates

ix

509

x 204.

CONTENTS

Hiroharu Sugawara, Tetsuya Hasegawa and Koichi Kitazawa Characterization of cuprate superconductors using tunneling spectra and scanning tunneling microscopy 563

Author index

609

Subject index

677

CONTENTS OF VOLUMES 1-30 V O L U M E 1: Metals 1978, 1st repr. 1982, 2nd repr. 1991; ISBN 0-444-85020-1 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

Z.B. Goldschrnidt, Atamic properties (free atom) 1 B.J. Beaudry and K.A. Gschneidner Jr, Preparation and basic properties of the rare earth metals' S.H. Liu, Electronic structure of rare earth metals 233 D.C. Koskenmaki and K.A. Gschneidner Jr, Cerium 337 L.J. Sundstr6m, Low temperature heat capacity of the rare earth metals 379 K.A. McEwen, Magnetic and transport properties of the rare earths 411 S.K. Sinha, Magnetic structures and inelastic neutron scattering: metals, alloys and compounds T.E. Scott, Elastic and mechanicalproperties 591 A. Jayaraman, High pressure studies: metals, alloys and compounds 707 C. Probst and J. Wittig, Superconductivity: metals, alloys and compounds 749 M.B. Maple, L.E. DeLong and B.C. Sales, Kondo efJect: alloys and compounds 797 M.P. Dariel, Diffusion in rare earth metals 847 Subject index 877

173

489

V O L U M E 2: Alloys and intermetallies 1979, 1st repr. 1982, 2nd repr. 1991; ISBN 0-444-85021-X 13. 14. 15. 16. 17. 18. 19. 20.

A. Iandelli anti A. Palenzona, Crystal chemistry ofintermetallic compounds 1 H.R. Kirchmayr and C.A. Poldy, Magnetic properties of intermetallic compounds" of rare earth metals 55 A.E. Clark, Magnetostrictive RFe2 intermetallic compounds 231 J.J. Rhyne, Amorphous magnetic rare earth alloys 259 E Fulde, Crystalfields 295 R.G. Barnes, NMR, EPR and M6ssbauer effect." metals, alloys and compounds 387 P. Wachter, Europium chalcogenides: EuO, EuS, EuSe and EuTe 507 A. Jayaraman, Valencechanges in compounds 575 Subject index 613

V O L U M E 3: Non-metallic compounds - I 1979, 1st repr. 1984; ISBN 0-444-85215-8 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

L.A. Haskin and T.E Paster, Geochemistry and mineralogy of the rare earths 1 J.E. Powell, Separation chemistry 81 C.K. Jorgensen, Theoretical chemistry of rare earths 111 W.T. Carnall, The absorption and fluoreseence spectra of rare earth ions in solution L.C. Thompson, Complexes 209 G.G. Libowitz and A.J. Maeland, Hydrides 299 L. Eyring, The binary rare earth oxides 337 D.J.M. Bevan and E. Summerville, Mixed rare earth oxides 401 C.P. Khattak and EEY. Wang, Perovsla'tes and garnets 525 L.H. Brixner, J.R. Barkley and W. Jeitschko, Rare earth molybdates (VI) 609 Subject index 655 xi

171

xii

CONTENTS OF VOLUMES 1 30

V O L U M E 4: Non-metallic c o m p o u n d s - II 1979, 1st repr. 1984; ISBN 0-444-85216-6 31. 32. 33. 34. 35. 36. 37A. 37B. 37C. 37D. 37E. 37F. 37G. 38. 39. 40.

1 J. Flahaut, Sulfides; selenides and tellurides J.M. Haschke, Halides 89 153 E Hulliger, Rare earthpnictides G. Blasse, Chemistry andphysies o f R-activated phosphors 237 M.J. Weber, Rare earth lasers' 275 EK. Fong, Nonradiative processes of rare-earth ions in crystals 317 J.W. O'Laughlin, Chemical spectrophotometric and polarographic methods' 341 359 S.R. Taylor, Trace element analysis o f rare earth elements' by spark source mass spectroscopy R.J. Conzemius, Analysis o f rare earth matrices by spark source mass spectrometry 377 E.L. DeKalb and V.A. Fassel, Optical atomic emission and absorption methods 405 A.P. D'Silva and VA. Fassel, X-ray excited optical luminescence o f the rare earths 441 EW.V Boynton, Neutron activation analysis 457 S. Schuhmann and J.A. Philpotts, Mass-spectrometric stable-isotope dilution analysis for lanthanides in geochemical materials 471 J. Reuben and G.A. Elgavish, Shift reagents and NMR o f paramagnetic lanthanide complexes 483 J. Reuben, Bioinorganic chemistry: lanthanides as probes in systems o f biological interest 515 T.J. Haley, Toxicity 553 Subject index 587

VOLUME 5 1982, 1st repr. 1984; ISBN 0-444-86375-3 41. 42. 43. 44. 45. 46.

M. Gasgnier, Rare earth alloys and compounds" as" thin films 1 E. Gratz and M.J. Zuckermann, Transport properties (electrical resitivity, thermoelectric power and thermal conductivity) o f rare earth intermetallic compounds 117 EP. Netzer and E. Bertel, Adsorption and catalysis on rare earth surJitces 217 C. Boulesteix, Defects' andphase transformation near room temperature in rare earth sesquioxides 321 O. Greis and J.M. Haschke, Rare earth fluorides 387 C.A. Morrison and R.P. Leavitt, Spectroscopic properties of triply ionized lanthanides in transparent host crystals 461 Subject index 693

VOLUME 6 1984; ISBN 0-444-86592-6 47. 48. 49. 50.

K.H.J. Buschow, Hydrogen absorption in intermetallic compounds' 1 E. Parth~ and ]3. Chabot, Crystal structures and crystal chemistry o f ternary rare earth-transition metal borides, silicides and homologues 113 P. Rogl, Phase equilibria in ternary and higher order systems with rare earth elements and boron 335 H.B. Kagan and J.L. Namy, Preparation o f divalent ytterbium and samarium derivatives and their use in organic chemistry 525 Subject index 567

VOLUME 7 1984; ISBN 0-444-86851-8 51. 52. 53.

P. Rogl, Phase equilibria in ternary and higher order systems with rare earth elements and silicon K.H.J. Buschow, Amorphous alloys 265 H. Schumann and W. Genthe, Organometallic compounds of the rare earths 446 Subject index 573

CONTENTS OF VOLUMES 1-30

xiii

VOLUME 8 1986; ISBN 0-444-86971-9 54. 55. 56. 57.

K.A. Gschneidner Jr and EW. Calderwood, Intra rare earth binary alloys: phase relationships, lattice parameters and systematics 1 X. Gao, Polarographic analysis of the rare earths 163 M. Leskel/i and L. Niinist6, Inorganic complex compounds I 203 J.R. Long, Implications in organic synthesis 335 Errata 375 Subject index

379

VOLUME 9 1987; ISBN 0-444-87045-8 58. 59. 60. 61.

R. Reisfeld and C.K. J~rgensen, Excited state phenomena in vitreous materials 1 L. Niinist6 and M. Leskelfi, Inorganic complex compounds II 91 J.-C.G. Btinzli, Complexes with synthetic ionophores 321 Zhiquan Shen and JuanOuyang, Rare earth coordination catalysis in stereospec~c polymerization Errata 429 Subject index 431

395

VOLUME 10: High energy spectroscopy 1988; ISBN 0-444-87063-6 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72.

Y. Baer and W.-D. Schneider, High-energy spectroscopy of lanthanide materials An overview 1 M. Campagna and EU. Hillebrecht, f-electron hybridization and dynamical screening of core holes in intermetallie compounds 75 O. Gunnarsson and K. Sch6nhammer, Many-bodyformulation ofspeetra ofmixedvalencesystems 103 A.J. Freeman, B.I. Min and M.R. Norman, Local density supercell theory ofphotoemission and inverse photoemission spectra 165 D.W Lynch and J.H. Weaver, Photoemission of Ce and its compounds 231 S. Hiifner, Photoemission in chalcogenides 301 J.E Herbst and J.W. Wilkins, Calculation of 4f excitation energies in the metals and relevance to mixed valence systems 321 B. Johansson and N. Mhrtensson, Thermodynamic aspects of 4f levels in metals' and compounds 361 EU. Hillebrecht and M. Campagna, Bremsstrahlung isoehromat spectlvseopy of alloys and mixed valent compounds 425 J. R6hler, X-ray absorption and emission spectra 453 ER Netzer and J.A.D. Matthew, Inelastic electron scattering measurements' 547 Subject index 601

VOLUME 11: Two-hundred-year impact of rare earths on science 1988; ISBN 0-444-87080-6 73. 74. 75. 76. 77. 78. 79.

H.J. Svec, Prologue 1 E Szabadvfiry, The history of the discovery and separation of the rare earths 33 B.R. Judd, Atomic theory and optical spectroscopy 81 C.K. Jorgensen, Influence of rare earths on ehemieal understanding and classification 197 J.J. Rhyne, Highlights Jkom the exotic phenomena of lanthanide magnetism 293 B. Bleaney, Magnetic resonance spectroscopy and hyperfine interactions 323 K.A. Gschneidner Jr and A.H. Daane, Physical metallurgy 409 S.R. Taylor and S.M. McLennan, The significance of the rare earths in geochemistry and cosmochemistry 485 Errata 579 Subject index 581

xiv

CONTENTS OF VOLUMES 1-30

V O L U M E 12 1989; ISBN 0-444-87105-5 80. 81. 82. 83. 84. 85. 86. 87.

J.S. Abell, Preparation and crystal growth of rare earth elements and intermetallic compounds" t Z. Fisk and LP. Remeika, Growth of single crystals from molten metalfluxes 53 E. Burzo and H.R. Kirchmayr, Physical properties of R2Fel4B-based alloys 71 A. Szytuta and J. Leciejewicz, Magnetic properties of ternary intermetallie compounds of the RTeX~ type 133 H. Maletta and W. Zinn, Spin glasses 213 J. van Zytveld, Liquid metals and alloys 357 M.S. Chandrasekharaiah and K.A. Gingerich, Thermodynamic properties o f gaseous species 409 W.M. Yen, Laserspeetroscopy 433 Subject index 479

V O L U M E 13 1990; ISBN 0-444-88547-1 88.

E.I. Gladyshevsky, O.I. Bodak and V.K. Pecharsky, Phase equilibria and crystal chemistry in ternary rare

89.

A.A. Eliseev and G.M. Kuzmichyeva, Phase equilibrium and crystal chemistry in ternary rare earth systems with chalcogenide elements 191 N. Kimizuka, E. Takayama-Muromachi and K. Siratori, The systems R203-MeOj M'O 283 R.S. Houk, Elemental analysis by atomic emission and mass" spectrometry with inductively coupled plasmas 385 P.H. Brown, A.H. Rathjen, R.D. Graham and D.E. Tribe, Rare earth elements in biologicalsystems 423 Errata 453 Subject index 455

earth systems with metallic elements

90. 91. 92.

1

V O L U M E 14 1991; ISBN 0-444-88743-1 93.

R. Osborn, S.W Lovesey, A.D. Taylor and E. Balcar, Intermultiplet transitions using neutron

spectroscopy 94. 95. 96. 97.

1

E. Dormann, NMR in intermetallic compounds 63 E. Zirngiebl and G. Gtintherodt, Light scattering in intermetallic compounds 163 P. Thalmeier and B. Lfithi, The electron-phonon interaction in intermetallic compounds N. Grewe and E Steglich, Heavyfermions 343 Subject index 475

225

V O L U M E 15 1991; ISBN 0-444-88966-3 98. 99. 100. 101. 102. 103. 104.

1 J.G. Sereni, Low-temperature behaviour of cerium compounds G.-y. Adachi, N. Imanaka and Zhang Fuzhong, Rare earth carbides 61 A. Simon, Hj. Mattausch, G.J. Miller, W. Bauhofer and R.K. Kremer, Metal-rich halides 191 R.M. Almeida, Fluoride glasses 287 K.L. Nash and J.C. Sullivan, Kinetics of complexation and redox reactions of the lanthanides in aqueous solutions 347 E.N. Rizkalla and G.R. Choppin, Hydration and hydrolysis oflanthanides 393 L.M. Vallarino, Macrocycle complexes of the lanthanide(III) yttrium(III) and dioxouranium(VI) ions from metal-templated syntheses 443 Errata 513 Subject index 515

CONTENTS OF VOLUMES 1-30

xv

MASTER INDEX, Vols. 1-15 1993; ISBN 0-444-89965-0

VOLUME 16 1993; ISBN 0-444-89782-8 105. 106. 107. 108. 109.

M. Loewenhaupt and K.H. Fischer, Valence-fluctuation and heavy-fermion 4fsystems I.A. Smirnov and V.S. Oskotski, Thermal conductivity of rare earth compounds' 107 225 M.A. Subramanian and A.W. Sleight, Rare earthspyrochlores R. Miyawaki and I. Nakai, Crystal structures of rare earth minerals 249 D.R. Chopra, Appearance potential speetroscopy of lanthanides and their intermetallics Author index 547 Subject index 579

519

VOLUME 17: Lanthanides/Actinides: P h y s i c s - I 1993; ISBN 0-444-81502-3 110. 111. 112. 113. 114. 115. 116. 117.

M.R. Norman and D.D. Koelling, Electronic structure, FelTnisurfaces, and supemonductivity inf electron metals' 1 S.H. Liu, Phenomenological approach to heavy-fermion systems 87 B. Johansson and M.S.S. Brooks, Theory of cohesion in rare earths and actinides 149 U. Benedict and W.B. Holzapfel, High-pressure studies Structural aspects 245 O. Vogt and K. Mattenberger, Magnetic measurements' on rare earth and actinide monopnictides and monochalcogenides 301 J.M. Fournier and E. Gratz, Transportproperties of rare earth and actinide intermetallics 409 W. Potzel, G.M. Kalvius and J. Gal, M6ssbauer studies' on electronic structure ofintermetallic compounds 539 G.H. Lander, Neutron elastic scattering from actinides and anomalous lanthanides 635 Author index 711 Subject index 753

VOLUME 18: Lanthanides/Actinides: Chemistry 1994; ISBN 0-444-81724-7 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129.

1 G.T. Seaborg, Origin of the actinide concept K. Balasubramanian, Relativistic effects' and electronic structure of lanthanide and actinide molecules 29 J.V. Beitz, Similaritiesanddi~rencesintrivalentlanthanide-andactinide-i~ns~luti~nabs~rpti~nspectra and luminescence studies 159 K.L. Nash, Separation chemistry for lanthanides and trivalent aetinides 197 L.R. Morss, Comparative thermochemical and oxidation reduction properties of lanthanides and actinides 239 J.W. Ward and J.M. Haschke, Comparison of4fand 5felement hydride properties 293 H.A. Eick, Lanthanide and actinide halides 365 R.G. Haire and L. Eyring, Comparisons of the binary oxides 413 S.A. Kinkead, K.D. Abney and T.A. O'Donnell, f-element speciation in strongly acidic media: lanthanide and mid-actinide metals, oxides',fluorides and oxide fluorides in superacids 507 E.N. Rizkalla and G.R. Choppin, Lanthanides and actinides hydration and hydrolysis 529 G.R. Choppin and E.N. Rizkalla, Solution chemistry ofactinides and lanthanides 559 591 J.R. Duffield, D.M. Taylor and D.R. Williams, The biochemistry of thef-elements Author index 623 Subject index 659

xvi

CONTENTS OF VOLUMES 1 30

VOLUME 19: Lanthanides/Actinides: P h y s i c s - II 1994; ISBN 0-444-82015-9 130.

E. Holland-Moritz and G.H. Lander, Neutron inelastic scattering from actinides and anomalous

131.

G. Aeppli and C. Broholm, Magnetic correlations in heavy-Jermion systems: neutron scattering from single crystals 123 P. Wachter, Intermediate valence and heavy fermions 177 J.D. Thompson and J.M. Lawrence, High pressure studies Physical properties of anomalous Ce, Yb and U compounds 383 C. Colinet and A. Pasturel, Thermodynamic properties of metallic systems 479 Author index 649 Subject index 693

lanthanides

132. 133. 134.

1

VOLUME 20 1995; ISBN 0-444-82014-0 135. 136. 137. 138.

Y. Ontdd and A. Hasegawa, Fermi surfaces ofintermetallic compounds 1 M. Gasgnier, The intricate world of rare earth thin films: metals, alloys, intermetallics, 105 chemical compounds,... E Vajda, Hydrogen in rare-earth metals, including RH2+x phases 207 293 D. Gignoux and D. Schmitt, Magnetic properties ofintermetallic compounds Author index 425 Subject index 457

V O L U M E 21 1995; ISBN 0-444-82178-3 139. 140. 141. t42. 143. 144. 145. 146.

R.G. Bautista, Separation chemistry 1 B.W. Hinton, Corrosion prevention and control 29 N.E. Ryan, High-temperature corrosion protection 93 T. Sakai, M. Matsuoka and C. Iwakura, Rare earth intermetallicsfor metal-hydrogen batteries 133 G.-y. Adacbi and N. Imanaka, Chemicalsensors 179 263 D. Garcia and M. Faucher, CrystalfieM in non-metallic (rare earth) compounds J.-C.G. Btinzli and A. Milicic-Tang, Solvation and anion interaction in organic solvents 305 V. Bhagavathy, T. Prasada Rao and A.D. Damodaran, Trace determination oflanthanides in high-purity rare-earth oxides 367 Author index 385 Subject index 411

V O L U M E 22 1996; ISBN 0-444-82288-7 147. 148.

C.E Flynn and M.B. Salamon, Synthesis and properties of single-crystal nanostructures 1 Z.S. Shan and D.J. Sellmyer, Nanoscale rare earth-transition metal multilayers: magnetic structure and

properties 149. 150. 151. 152.

81

W. Suski, The ThMn12-type compounds of rare earths and actinides: structure, magnetic and related properties 143 L.K. Aminov, B.Z. Malkin and M.A. Teplov, Magnetic properties of nonmetallic lanthanide compounds 295 E Auzel, Coherent emission in rare-earth materials 507 M. DolgandH. Stoll, Electronicstructurecalculationsformoleculescontaininglanthanideatoms 607 Author index 73 l Subject index 777

CONTENTS OF VOLUMES 1-30

xvii

V O L U M E 23 1996; ISBN 0-444-82507-X 153. 154. 155. 156. 157. 158.

J.H. Forsberg, NMR studies of paramagnetic lanthanide complexes' and shift reagents 1 N. Sabbatini, M. Guardigli and I. Manet, Antenna effect in encapsulation complexes oflanthanide ions 69 121 C. GSrller-Walrand and K. Binnemans, Rationalization ofcrystal-fieldparametrization Yu. Kuz'ma and S. Chykhrij, Phosphides 285 S. Boghosian and G.N. Papatheodorou, Halide vapors' and vapor complexes 435 R.H. Byrne and E.R. Sholkovitz, Marine chemistry and geochemistry of the lanthanides 497 Author index 595 Subject index 631

V O L U M E 24 1997; ISBN 0-444-82607-6 159. 160. 161. 162. 163. 164.

P.A. Dowben, D.N. Mcllroy and Dongqi Li, SurJhce magnetism of the lanthanides 1 RG. McCormick, Mechanical alloying and mechanically induced chemical reactions 47 A. Inoue, Amorphous, quasierystalline and nanocrystalline alloys in AL and Mg-based systems 83 B. Elschner and A. Loidl, Electron-spin resonance on localized magnetic moments in metals 221 N.H. Duc, [ntersublattice exchange coupling in the lanthanide-transition metal intermetallics 339 R.V. Skolozdra, Stannides of rare-earth and transition metals 399 Author index 519 Subject index 559

V O L U M E 25 1998; ISBN 0-444-82871-0 165. 166. 167. 168.

H. Nagai, Rare earths in steels 1 R. Marchand, Ternary and higher order nitride materials 51 C. GSrller-Walrand and K. Binnemans, Spectral intensities o f f - f transitions G. Bombieri and G. Paolucci, Organometallic zr complexes" of the f-elements Author Index 415 Subject Index 459

101 265

V O L U M E 26 1999; ISBN 0-444-50815-1 169. 170. 171. 172.

D.E McMorrow, D. Gibbs and J. Bohr, X-ray scattering studies oflanthanide magnetism 1 A.M. Tishin, Yu.I. SpichkJnandJ. Bohr, Static and dynamic stresses 87 N.H. Duc and T. Goto, Itinerant electron metamagnetism of Co sublattice in the lanthanide-cobalt intermetallics 177 A.J. Arko, RS. Riseborough, A.B. Andrews, J.J. Joyce, A.N. Tahvildar-Zadeh and M. Jarrell, Photoelectron spectroscopy in heavy fermion systems: Emphasis on single crystals 265 Author index 383 Subject index 405

V O L U M E 27 1999; ISBN 0-444-50342-0 173. 174. 175.

P.S. Salamakha, O.L. Sologub and O.I. Bodak, Ternary rare-earth germanium systems 1 P.S. Salamakha, Crystal structures' and crystal chemistry of ternary rare-earth germanides 225 B.Ya. Kotur and E. Gratz, Scandium alloy systems and intermetallics 339 Author index 535 Subject index 553

xviii

CONTENTS OF VOLUMES 1-30

V O L U M E 28 2000; ISBN 0-444-50346-3 176. 177. 178. 179. 180.

J.-R Connerade and R.C. Karnatak, Electronic excitation in atomic species 1 G. Meyer and M.S. Wiekleder, Simple and complex halides 53 R.V Kumar and H. Iwahara, Solid electrolytes 131 A. Halperin, Activated thermoluminescence (TL) dosimeters and related radiation detectors 187 K.L. Nash and M.R Jensen, Analytical separations of the lanthanides: basic chemistry and

methods

311

Author index Subject index

373 401

V O L U M E 29: The role of rare earths in catalysis 2000; ISBN 0-444-50472-9 181. 182. 183. 184. 185. 186.

1 P. Maestro, Foreword V Paul-Boncour, L. Hilaire and A. Percheron-Gu~gan, The metals' and alloys' in catalysis 5 H. Imamura, The metals and alloys (prepared utilizing liquid ammonia solutions) in catalysis II M.A. Ulla and E.A. Lombardo, The mixed oxides 75 J. Ka~par, M. Graziani and P. Fomasiero, Ceria-containing three-way catalysts 159 A. Corma and J.M. L6pez Nieto, The use of rare-earth-containing zeolite catalysts 269 S. Kobayashi, Triflates 315 Author index 377 Subject index 409

45

V O L U M E 30: High-Temperature Superconductors - I 2000; ISBN 0-444-50528 8 187. 188. 189. 190. 191. 192. 193. 194.

M.B. Maple, High-temperature superconductivity in layered cuprates: overview 1 B. Raveau, C. Michel and M. Hervieu, Crystal chemistry of superconducting rare-earth cuprates 31 Y. Shiohara and E.A. Goodilin, Single-crystal growth for science and technology 67 E Karen and A. Kjekshus, Phase diagrams and thermodynamic properties 229 B. Elschner and A. Loidl, Electron paramagnetic resonance in cuprate superconductors and in parent compounds 375 417 A.A. Manuel, Positron annihilation in high-temperature superconductors W.E. Pickett and I.I. Mazin, RBa2Cu307 compounds: electronic theory and physical properties 453 U. Stanb and L. Soderholm, Electronic 4fstate splittings in cuprates 491 Author index 547 Subject index 621

Handbook on the Physics and Chemistry of Rare Earths Vol. 31 edited by K.A. Gschneidner, Jr, L. Eyring and M.B. Maple © 2001 Elsevier Science B.V. All rights reserved

Chapter 195 OXYGEN NONSTOICHIOMETRY IN YBa2CuaOx

AND LATTICE EFFECTS

Phase Transitions, Structural Distortions and Phase Separation Emanuel KALDIS i.R. Laboratorium fiir Festk6rperphysik ETH, 8093 Zurich, Switzerland, [email protected], [email protected]

Comems List of abbreviations 1. Introduction 2. Average crystallographic structure 3. Structural evolution of 123-O~ with nonstoichiometry 3.1. Synthesis, oxygen and carbonate control 3.1.1. Nonequilibrium synthesis, quenching 3.1.2. Synthesis near thermodynamic equilibrium 3.1.2.1. Overdoped samples, x > 6.95 3.l.2.2. Optimally doped, and underdoped samples, x < 6.95 3.1.3. High-accuracy determination of the oxygen content 3.1.4. Mass-spectrometric determination of the carbonate content 3.2. Changes of the lattice parameters and bond lengths with the oxygen nonstoichiometry: the structural T ~ O transformation 3.2.1. Nonequilibrium samples, mainly underdoped regime 3.2.1.1. Lattice parameters of non-equilibrium samples 3.2.1.2. C ~ O bonds 3.2.1.3. B ~ O and Y - O bonds 3.2.1.4. Application of the Bond Valence Sum (BVS) method to 123-O, I 3.2.1.5. Mobility of oxygen, reordering at quenching

2 3 8 11 11 12 13 14

15 17 20

22 22 22 25 29

30 33

3.2.2. Equilibrium samples with highresolution oxygcn determination 3.2.2.1. Lattice parameters vs. oxygen content 3.2.2.2. X-ray diffractometry, T=300K 3.2.2.3. Lattice parameters of samples with Ba carbonate precursor (CAR samples) 3.2.2.4. Lattice parameters of BAO and DO samples 3.3. The T --+ O transition 3.3.1. The structural T - * O transformation at the insulator superconductor transition 3.3.2. Raman investigations near the T-O transition 4. Local structure 4.1. Anomalies of the Debye~Waller factor: the split atom model of the chains 4.2. Lattice distortions: investigations of the pair distribution function (PDF) 4.3. Distortion of the planes: high-resolution electron diffraction 4.3.1. Nanoscale phase separation. The Phillips model 4.4. EXAFS and PDF investigations of the apical oxygen 5. Insulating, underdoped, optimally doped phases: phase diagram, phase separation and superstructures 5.1. Phase diagram and superstructures: experiments and calculations

38 38 38

39 44 51

58 62 64 65 67 68 71 72

79 79

2

E. KALDIS

5.1.1. Experimental investigations: EM, XRD and ND 5.2. Model calculations 5.3. Tc dependence on nonstoichiomett3, and hole concentration: applications of the BVS method - II 5.4. Pressure dependence of Tc as a function of nonstoichiometry: giant dTo/dP effects at x ~ 6.75 and x ~ 6.4 5.4.1. Uniaxial pressure investigations dTc/dP i as a function of nonstoichiometry 5.4.2. The structural, optical and magnetic anomalies at x ~ 6.75 5.5. Raman scattering as a function of nonstoichiometry I 5.5.1. The apex Raman phonon and phase separation 5.5.2. The steps in the in-phase phonon frequency 6. Overdoped phase - I: displacive martensitic transformation 6.1. Overdoped phase investigated with equilibrium samples and high resolution volumetric analysis 6.2. Dimpling (Cu2-O2,O3) transition at the onset of the overdoped regime: neutron diffraction 6.3. Raman scattering as a function of nonstoichiometry - II. Softening of the in-phase 0 2 , 0 3 phonon 6.4. Local structure of the planes: EXAFS in the optimally doped and overdoped regimes

79 87

93

99

106 108 110 110 114 117

117

118

119

120

6.5. The displacive martensitic transformation in the overdoped regime of 123-Ox 7. Overdoped phase - II: phase separation in 123-OX 7.1. The splitting of the diamagnetic transition in the overdoped regime 7.2. The shell model 7.2.1. Oxygen diffusion in the grains: no surface barrier 7.3. The splitting of the diamagnetic transition as a function of the exact oxygen content: indication for phase separation 7.4. Phase separation due to stripe formation: separation of spin and charge 8. Overdoped phase - I I I : phase separation in the Ca-overdoped regime, Y(Ca)-123-O x 8.1. Oxygen content: the A and B states 8.2. Structural, optical and magnetic characterization of Ca-123-Ox 8.2.1. Structural characterization 8.2.2. Raman characterization 8.2.3. Magnetic characterization 8.3. EXAFS investigations 8.3.1. The A - B transitions and ageing effects 8.4. Ca doping and localized carriers 9. Significant points of a doping-dependent phase diagram 123-Ox (6.0 < x < 7.0) 9.1. A T ~ phase diagram for 123-Ox Summary Acknowledgments References

127 131 131 133 135

138 141 145 146 151 151 154 156 159 161 163 165 171 174 176 177

List of abbreviations 21) 3D AF

Two-dimensional Three-dimensional Anti ferromagnetic

CVM DO

Clusters Variation Method Synthesis of 123-Q using Ba metal as Ba source (directly oxidized) Difference Pair Density Function Electron diffraction

ASYNNNI Asymmetric Next-Nearest Neighbor Ising model BAO Synthesis of 123-O, using BaO as Ba source, by previously expelling the carbonate of BaCO 3 with high temperature annealing

DPDF El) EM EXAFS

ElectTon energy loss spectroscopy Electron microscopy Extended X-ray absorption fine structure

BVS

Bond Valence Sum method

FWHM

Full width at half maximum

CAR

Synthesis of 123-O~ using BaCO3 as Ba source

EELS

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN Y B a 2 C u 3 0 ~. HREM HT~ HToS I-M LT LO MS ND NEXAFS nh

NN NPD O Op Ortho-I Ortho-II

High-resolution electron microscopy High critical temperature High critical temperature superconductivity Insulator-metal transition Low temperature Longitudinal optical phonon Multiple backscattering Neutron diffraction Near edge X-ray absorption fine structure positive superconductingcarriers (holes) Nearest neighbors Neutron powder diffraction Orthorhombic phase Oxygen of the planes (02, 03) Orthurhombic phase of 123-Ox Superstructure of the orthorhombic phase of 123-Ox, with only every second chain occupied

PDF PICTM Raman RT SQUID SS T-O Tc Tc,m.x T~.o.~ot XPS XRD

3

Pair distribution function Pressure-induced charge transfer model Raman spectroscopy Room temperature Superconducting quantum interference device Single backscattering TetragonaNOrthorhombic transition Critical temperature of superconductivity Maximal temperature of diamagnetic transition as a function of doping Onset temperature of diamagnetic transition X-ray photoelectron spectroscopy X-ray diffraction

1. Introduction A very fruitful approach to investigate the properties o f a solid is the variation o f the carrier concentration. This can be achieved either by doping with a foreign atom or, for "bertholidic" compounds, by manipulating the nonstoichiometry. In nonstoichiometric solids variation o f the stoichiometric ratio leads to an intrinsic change of doping and allows the synthesis o f a series o f samples with different carrier and/or defect concentrations. The investigation o f the structural, chemical and physical properties o f such series allows the construction o f the various pseudo-binary phase diagrams (thermodynamic, structural, magnetic etc.) as a function o f intrinsic (and extrinsic) doping. In view o f the almost-always existing interaction between the structural and physical properties, this integrated approach leads to the most balanced picture of the properties o f a solid, showing clearly their evolution and their interactions with the change o f the carriers. On the contrary, the investigation of only a few "representative" compositions, very popular in the literature because it saves time, may lead, in the case of complex materials, to misinterpretations. In such cases, overseeing the existence o f superstructures, other homologous phases, miscibility gaps, phase transitions, etc., can lead to substantial errors. In particular, the homogeneity o f large nonstoichiometric regions should be questioned in that respect. An example from solid-state chemistry is the phase diagram o f the praseodymium-oxygen system whose homogeneity ranges were found with HREM to consist of a very large number o f hitherto unknown phases (Eyring 1979). An example

4

E. KALDIS

3901 T, K

270~ 230 190 t50

ZO

2.2

2.4

2.6

2.8

3.0

Fig. 1. Subsolidus T-x phase diagram of the LaH2 (metallic)-LaH3 (semiconducting) system based on differential scanning calorimetry and X-ray diffraction measurements. Instead of the originally assumed continuous solid solution with cubic Fm3m structure, the metal to semiconductortransition proceeds via at least 9 phases separatedby miscibility gaps. After Conder et al. (1991). from solid-state physics of the materials with valence instabilities is the TmSex phase (NaC1 structure) of the T m - S e system whose large homogeneity range was found to consist of several phases with different Tm valence (Kaldis and Fritzler 1982). Another intimidating example from the physics of the hydrides is the compositionally driven metal-semiconductor transition in the hydrides of lanthanum LaH2-LaH3 and cerium CeH2-CeH3 (fig. 1) which have been shown to consist of up to 9 subsolidus phases separated by miscibility gaps (Conder et al. 1991)! Up to now the physical properties of these new hydride phases have not been completely investigated due to their extreme sensitivity to contamination. YBa2Cu3Ox (123-Ox), ingeniously discovered by Wu et al. (1987) shortly after the historical opening of the HTc era by Bednorz and Mfiller (1986), is a nonstoichiometric solid par excellence: its oxygen content x varies between 6.0 and 6.990, and dramatically affects its properties. As we will discuss later, the oxygen nonstoichiometry of 123 results from the less stable square planar coordination of the Cul in the chains. With increasing O content the tetragonal antiferromagnetic insulator YBa2Cu306 (123-O6) undergoes at x ~ 6.30 a structural transformation to orthorhombic structure and at x ~ 6.38 to a superconductor with low Tc. Further increase of the carrier concentration by O-doping leads to an increase of To, through the underdoped (x <<.6.90), up to the optimally doped phase with 92.5 K (6.90 ~<x ~< 6.95). In the narrow ouerdoped phase (6.95 ~<x ~ 7.00) Tc decreases to ~88 K. The nonlinear change of T~ with oxygen content (fig. 2) indicates the possibility of the existence of complex phase relationships in 123-O~ associated with changing superconducting interactions. In fact this fascinating compound and the rest of

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30~

6,4 O0 ~ ,

90i " E 80 i a_ o I-70. I--

0

60 ¸ 50 40 30

6,5

6,6

I

I

'

6,7 6,8 6,9 7,0 '1' ' underdo ,ed . . ; . . . ~ . I

I

Conder et al. 1994 Zechetal. 1995 " -.-•

"0

i~

p

-~

~

.-

~

o

0 o

~-!

"

)

'

~ I~--~.1

i o

"f''•

o ( o ~"

'

o

~i ~ ,

6,4

,

,

i

,

,

,

,

6,5

E ,,

50

[ .Lo

6,7

Tcl -40 ,

1

6,6

80

60

o ,

90

70

x

"o

=oo

100

~1

'

O

~

'

6,8

6,9

I

30

7,0

Oxygen content x Fig. 2. Tc and Tc,,,n~et as a function of the oxygen nonstoichiometry. The nonlinear dependence of T, due to changing superconducting interactions indicates the existence of complex phase relationships. The Tc and Tc,o~et vs. x curves have clear differences. Note the abrupt step in Tc, onset a t x ~ 6.75-6.80, indicating a miscibility gap (hatched) consistent with fig. 28. After data of Conder et al. (1994a) and Zech et al. (1995a,b).

the HTc cuprate superconductors are among the most complex solids investigated up to now. As some of this complexity is revealed only with sophisticated methods of synthesis and characterization, it is no wonder that 14 years after the great discovery of Bednorz and Miiller (1986) there is an appreciable gap between theoretical models and experimental results. Some of the reasons for the existence of this gap lie in the materials. With the exception of discovering new structures and their homologues, research in the controlled synthesis of the known materials in order to understand their chemical and thermodynamic properties, their numerous deviations from the average structure, their instabilities and metastabilities, the mechanism of defect formation and the structure-properties relationships have been scarcely funded, being considered as not of primary importance to superconductivity. The lack of such profound knowledge about the materials combined with low accuracy of the oxygen determination or in most cases only "estimations" of it, and the investigation each time of a small number of "representative" compositions obscured many important but subtle trends. HTc superconductors have a stability range squeezed between the insulator and metal regimes and are very near to the border of thermodynamic instability, as is

6

E. KALDIS

already visible from the properties of the chains in 123-Ox. Subtle changes of their structural properties result from changes in the superconducting properties, which are easy to overlook, if painstaking work is avoided. Other difficulties concerning some single-crystalline samples are the time-consuming diffusion of oxygen in macroscopic crystals, the difficulty to anneal and control defects in larger single crystals, the difficulty to use high-resolution oxygen analysis, and their impurity content from flux and crucible. Based on the above, we may, therefore, estimate that more than 70% of the work in HTc superconductivity up to now has been performed with samples having several of the above drawbacks. The situation becomes much more complex if we take into consideration that some of the extraordinary difficulties concerning the preparation and characterization of HTc materials are due to their intrinsic inhomogeneities (see, e.g., Bordet et al. 1993). This is a completely new characteristic for solid-state materials, opposing all precedents. For five decades, materials research, inspired from the properties of semiconductors, had as ultimate goal the ideally homogeneous and structurally perfect material. On the other hand, with progressing time the importance of intrinsic inhomogeneities for superconductivity is increasingly recognized. The ordered charge and spin separation leading to stripes (Tranquada et al. 1995) is an intrinsic inhomogeneity par excellence. Most structural instabilities and thermodynamic metastabilities are not due to "bad samples" but are closely related to the anomalous properties of these strange metals. Thus, we have to differentiate much more carefully than in other materials between intrinsic and extrinsic ("bad samples") inhomogeneities. Figure 3 illustrates this aspect. The width of the diamagnetic transition is generally considered to be a very sensitive criterion of high-quality crystals with very good homogeneity. The figure shows a crystal with ATo = 300 inK! Nevertheless it still has several domains with oxygen inhomogeneities (different lengths of c-axis). Probably this is an extrinsic effect, but how many crystals have been investigated with that method? Additional complexity is due to the very small correlation length of the HT~ superconductors. This induces a great sensitivity to nanoscale inhomogeneities, but also lessens sensitivity to grain size reductions, segregations and precipitations down to mesoscopic scale. This extraordinary physical property has, however, also an advantage for materials research in general as it can be used as a probe to investigate for the first time phase relationships in a very small length scale. This is another important - but much less k n o w n - advantage of the discovery of Bednorz and Mtiller (1986), which may widen our understanding of complex materials in the future. As most properties of 123-Ox change - some of them, like T~, dramatically - with the oxygen content, a complete review of the changes of properties as a function of the nonstoichiometry would occupy the space of a large monograph. In this review we will focus mainly on the complexity of the structural phenomena and of the phase relationships of 123-Ox, as well as on some of their consequences for the physical properties. Particular attention will be given to lattice-parameter anomalies, the local structure, and some hitherto unknown phase transitions triggered by doping. Using slowly cooled samples and high-resolution oxygen determination some of these anomalies are becoming visible

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa~Cu30 x

20

6.

(7 - 8)

I/°.oo O.os

Fig. 3. Oxygen content 7 - 6 in the (z,y) plane of a "high-quality" single crystal of 123-O~ (Tc ~ 93 K, ATc < 300 inK!). Investigations with very high energy syncbaotron radiation allow the measurement of minute changes of the c-axis throughout the bulk of the crystal, and the detection of the corresponding changes due to inhomogeneous oxygen distribution. The plot shows that this crystal contains bands of different c-axis lengths along the z-direction. After Qadri et al. (1997).

even by diffraction methods. Investigations by mesoscopic length scale methods (EXAFS, micro-Raman) showphase separation phenomena, which are supported by the splitting of the diamagnetic transition, and complex structural changes in the as yet less investigated overdoped region. This is the reason why an appreciable part of this review will focus on the overdoped region (oxygen and Ca doping). It is hoped that these new results may contribute towards moving the focus of many theoretical models away from the ideal lattice and homogeneous phases. The simple strategy followed here is to discuss structural anomalies appearing at certain oxygen contents and scout the field of physical measurements for changes of properties at the same oxygen content. The opposite is useful as well: looking for structural anomalies where changes in the properties appear. Both approaches can also be used for proposing the existence of new phases to be confirmed (or not) by later investigations. Based on structural, magnetic and Raman investigations discussed in this review, we will propose a tentative T-x phase diagram in sect. 9. As the literature of 123-Ox is extremely large, it is outside the scope of this review to give complete references. Thus, many but not all papers related to the aspects discussed in this review are given in the list of references (as of late November 2000). Some facets of the problem which have been discussed in detail in past reviews will be discussed only briefly here, or references will be given. This includes, e.g., a large part of the diffraction crystallography of single crystals, inelastic neutron scattering, and NMR investigations.

8

E. KALDIS

c

./

b

Fig. 4. Schematic drawing illustrating the two different coordinations of Cu (chains and planes) due to the existence of two vacant oxygen sites, and the tripling of the perovskite cell of YBa2Cu307. After Beyers and Shaw (1989).

2. Average crystallographic structure The atomic positions of the "real crystal" are to some degree different from the atomic positions of the "ideal crystal" given by the space group (see, e.g., Kaldis 1994). The structure used to describe the real crystal is the "average structure" and is only an approximation of the real atomic structure (local structure). We will discuss the local structure in sect. 4 and some subsequent sections. Here we give the highlights of the average structure, as detailed descriptions have been given in previous reviews. Detailed references of the discovery of Wu et al. (1987) and the early studies of the 123-O~ structure are given, e.g., in the first review on this subject (Beyers and Shaw 1989). The reviews in the series edited by Ginsberg (1990-95) and Narlikar (since 1990), as well as the most recent review of Radaelli (1998) are recommended reading. YBazCu3Ox has a structure derived from the cubic perovskite structure ABO3. in this structure the larger cations A occupy the center of the cage formed by corner-sharing oxygen octahedra and the smaller B cations occupy the centers of the oxygen octahedra, see e.g., Goodenough and Longo (1970). The smaller cation is Cu here, but there are two larger cations Y and Ba occupying the A sites. Due to ordering of the two Ba and the one Y cations the unit cell is tripled with the sequence B a - Y - B a of the subcells (fig. 4). Ideally such a perovskite structure contains 3 x 3 = 9 oxygen atom sites. It was found, however, that the 123 unit cell contains only 7 oxygen atoms. The vacant oxygen sites are in the plane of Y (oxygen coordination 8, slightly distorted square prism) and in the basal copper plane (Ba oxygen coordination 10). In this way two different Cu sites result, forming the chains and planes of the 123 structure:

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30~

0(2)

cu(2)

Cu(I]

9

Fig. 5. Unit cell of the average structure of the orthorhombic YBa~Cu307 (Pmmm space group) with labels of the various atom sites. The 05 sites are shown empty. The (almost) completely oxygen-depleted Cul-(O5)-Cul chains form the a-axis and the full Cul-O4-Cul chains form the b-axis. The c-axis is along the Cul-O1-Cu2-Cu2-Ol-Cul strings.The five-fold oxygen coordination of the Cu2 and the four-fold coordination of Cul are clearly seen. Note that the superconducting Cu2-O2,O3 planes are not flat. (cf. fig. 7).

- The C u l occupy the corners o f the basal plane and have a square planar coordination o f oxygen, forming CuO2 chains parallel to the b-axis (fig. 4). The oxygen vacancies are along the a-axis between the C u l o f neighboring chains ( 0 5 or antichain site), and as we will see they may also be, under certain conditions, partly occupied (fig. 34). The occupied oxygen sites near C u l (fig. 5) are the chain oxygen 0 4 (along the b-axis) and the apex oxygen O11 along the c-axis. Due to this asymmetric occupancy o f the a- and b-axis the latter is longer (1.8%) than the a-axis, leading to an orthorhombic structure near x ~ 6.4, crystallizing in the centrosymmetric P m m m space group. At lower oxygen contents the structure is tetragonal (P4/mmm) (fig. 6). This leads to a tetragonal-orthorhombic ( T - O ) phase transition with increasing oxygen content, near the onset o f superconductivity (sect. 3.3). - The Cu2 sites have afivefoldpyramidal oxygen coordination, forming CuO5 pyramids with apex the bridging O1 and basis the 0 2 and 0 3 oxygens. Cu2 and 0 2 , 0 3 form the superconducting Cu202 planes. These are not flat planes, because a deformation (Capponi et al. 1987), dimpling, exists along the c-axis, the 0 2 and 0 3 lying above Cu2 at different heights ( 0 3 higher than 0 2 ) (fig. 7). The O 2 - C u 2 - O 2 zig-zag chains lie along the a-axis and the O 3 - C u 2 - O 3 along the b-axis. Usually the C u 2 0 ; plane is considered as one block layer. However, as neutron diffraction (Conder et al. 1994a; see also sect. 6.1, fig. 66) and E X A F S (R6hler et al. 1997a,b, Kaldis et al. 1997b) have shown that the separation between Cu2 and 0 2 , 0 3 changes sensitively with the carrier concentration, one may consider the Cu202 planes as consisting o f three stacked layers: Cu2 and the double 0 2 / 0 3 . We note that, as shown in 1 Throughout this review we use the O1 label for the apical oxygen site and 04 for for the chain oxygen. As the reverse notation for the chain and apex oxygen sites often occurs in the literature, some authors call the apical oxygen "bridging oxygen" to avoid misunderstandings. In fact, an important function of this oxygen is that it provides a bridge between the Cul system of the chains and the Cu2 system of the planes.

10

E. KALDIS

Fig. 7. View of the Cu202 superconducting plane to illustrate dimpling. At the front the Cu2-O2-Cu2 zig-zag chains can be seen (ll a-axis). The Cu2-O3-Cu2 zig-zag chains are II b-axis. As a result of the different lengths of the Cu2-O3 and Cu2-O2 bonds a corrugation exists along the Cu2-Cu2 (110) chains - lying lower (cf. figs. 15 and 70). Also the 0 2 - 0 3 - 0 2 (100) strings have a modulation with 02 lying higher than 03. Black spheres, Cu, white spheres, O. On the left-hand side the basis of one Cu2-O5 pyramid is shown (single solid bond lines). Fig. 6. Unit cell of the average structure of the tetragonal YBa2Cu306 (P4/m mm space group) with oxygendepleted C u l - C u t chains along the b-axis.

Table 1 Structural parameters (~, units) in highly overdoped YBa2Cu30., with x=6.994, measured with neutron diffraction refmement, at T = 1.5 K (ILL-Grenoble, DIA diffractometer)"

Cu2-O 1 - 2.268(2)

Cu2-O2 = 1.9265(4)

Cul-O1 = 1.857(2)

Cul 04 = 1.9407(0)

5 ( 0 2 = 2.403(1)

Y O3=2.384(1)

B ~ O 2 = 2.988(2) a [AI = 3.811548(5),

Cu2-O3 = 1.9589(4)

Cu2-Cu2

3.396(2)

B ~ O 3 - 2.954(2)

Ba O1=2.7342(3)

B ~ O 4 = 2.862(2)

b [A] = 3.88137(7),

c [~,] = 11.6477(2),

V [~3] = 172.32(1).

This is the highest oxygen content measured with the high-resolution method described in sect. 3.1.3. The sample was prepared similar to the CAR samples (sect. 3.1.2). The synthesis conditions are described in sect. 8. As usual in neutron diffraction the standard deviations of the cell parameters do not include the uncertainty in the effective neutron wavelength 3, = 1.90797 A. After B6ttger et al. (1996).

-

-

fig. 5, t w o n e i g h b o r i n g C u 2 0 2 p l a n e s are s e p a r a t e d b y t h e Y-layer. T h e C u 2 0 2 - Y - C u 2 0 2 t r i p l e l a y e r s are s e p a r a t e d b y t h e c h a i n s . T a b l e 1 g i v e s s e l e c t e d b o n d l e n g t h s f o r t h e 123-Ox s t r u c t u r e . A p p r e c i a b l e d i f f e r e n c e s exist among the various Cu-O bond lengths: T h e C u l - O 4 b o n d l e n g t h a l o n g t h e c h a i n ( b - a x i s ) is 1 . 9 4 0 7 ( 0 ) A . T h e b o n d o f C u l to t h e a p i c a l o x y g e n , C u l - O 1 , at 1 . 8 5 7 ( 2 ) * is t h e s h o r t e s t C u - O b o n d i n t h e s t r u c t u r e a n d is t h e o n e w h i c h c h a n g e s s t r o n g l y w i t h i n c r e a s i n g o x y g e n n o n s t o i c h i o m e t r y (sect. 3.2.1.2).

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN YBazCu30x

11

- In contrast, the Cu2 bond to apex oxygen is, with 2.268(2) A, the largest of the structure. Its length also varies with stoichiometry, and influences the degree of coupling between the chains and the planes. - The Cu2 bonds to the 02, 03 oxygens of the planes (at the basis of the CuO5 pyramid) are not equal, Cu2-O2 = 1.9265(4)-A, Cu2-O3 = 1.9589(4)A, leading to O - C u 2 - O angles smaller than 180 °. This destroys the planar character of the Cu2 planes and introduces the above-mentioned dimpling. As we will see, dimpling is dependent on the oxygen nonstoichiometry and scales with To (sect. 6). Although hitherto rather neglected, as we will see later it is an important parameter of HTc superconductivity. In addition to the deviation from planarity of the C u 2 Q planes, a zig-zag modulation of the chains has been found due to large anisotropic thermal vibrations perpendicular to the b-axis (Capponi et al. 1987, Francois et al. 1988). This is in agreement with the thermal instability of the chains which start losing oxygen at T > 350°C, as will be discussed in sect. 4.1. The diffusion coefficient of the various oxygen sites was investigated (isotope exchange) in detail by Conder et al. (1994b) and Conder (2000), cf. sect. 7.2.1. 3. Structural evolution of 123-0x with nonstoichiometry The nonstoichiometry of 123-Ox became apparent very early from the differences in structural and superconducting properties of samples synthesized under various conditions. The large thermal ellipsoids of the structure refinements (Capponi et al. 1987) indicated also the relative instability of the oxygens of the chains (O4), resulting from the less stable bonding of the square planar coordination of Cul. At the same time it was found that T~ changes with nonstoichiometry. The chargetransfer model gave a first insight to this dependence: The superconductivity in the Cu202 planes depends on the charge transfer between the chains and the planes (Miceli et al. 1988, Cava et al. 1988a,b) and this increases with the occupation of the 04 sites of the chains as the apical bond decreases (sect. 3.2.1.2). Thus, nonstoichiometry became an instrument to vary the properties of 123-Ox from the tetragonal antiferromagnetic insulator 123-O6.4 to the superconductor 123-O6.9 with Tc ~ 92 K! Since then, structural and thermodynamic investigations as a function of the oxygen nonstoichiometry have become very important in understanding some aspects of HTc superconductivity. The lability of the C u l - O 4 bond, responsible for the oxygen nonstoichiometry, also allows ordering processes in the chains. Indeed, it was found very soon that Tc depends not only on the occupancy of the chains but also on the ordering imposed by the thermal history (sect. 3.2.1.5), pressure (sect. 5.4), chemical history (sect. 3.2.2.5) and the general conditions of synthesis of the samples. Thus, the appreciable dependence of structure on oxygen content varies considerably with synthesis conditions. With a lot of fantasy and mostly no adherence to equilibrium thermodynamics many variations of the synthesis conditions have been tried. 3.1. Synthesis, oxygen and carbonate control Independently of the advantages and disadvantages of the synthesis method used, a great disadvantage of many publications in the literature concerned with changes of various

12

E. KALDIS

properties as a function of nonstoichiometry is the relatively small number of investigated compositions. The complex perovskite structure of the HTc cuprates is a sensitive probe of the changes of various physical parameters. As some of these changes are abrupt and rather small, one needs a large number of investigated compositions and a highly accurate determination of the oxygen content in order to see the trends and be able to detect these effects. Both have been missing in the mainstream literature so that, e.g., some structural aspects of the overdoped regime could not be systematically investigated. Thus, the discussion of the structural properties of nonequilibrium samples (sect. 3.2.1) deals with the underdoped phase. Several properties of the overdoped phase have been studied with equilibrium samples and will be discussed in sections 3.2.2, 6, 7 and 8.0. 3.1.1. Nonequilibrium synthesis, quenching To reach low oxygen contents (tetragonal samples) at rather low temperatures, low oxygen partial pressures Po2 are necessary. As it is rather difficult to reach a good control of low Po2, most researchers have used a "practical" alternative: reduction of oxygen by annealing at high temperatures followed by quenching. This leaves the open question of how the extremely sensitive defects react, e.g., to the strains resulting from this rather brutal procedure. In view of the practical impossibility to reach complete quenching, which soon enough became clear (e.g., Jorgensen et al. 1990a), it is rather astonishing that a large majority of researchers followed this technique. Thus, a large number of laboratories prepared nonstoichiometric samples by quenching them rapidly from high temperatures, 600-1000°C (Jorgensen et al. 1987a, TakayamaMuromachi et al. 1987, Farneth et al. 1988, among others) to liquid nitrogen or room temperatures. According to an in-situ neutron diffraction investigation (Jorgensen et al. 1987b) quenching from high temperatures crosses the phase-transition line tetragonal ---+ orthorhombic (T---+ O) and, therefore, increases the degree of disorder of the samples. Thus, e.g., these samples show a rather smooth variation of the dependence of Tc on oxygen content and lose superconductivity at x ~ 6.5. Variation of the oxygen stoichiometry at lower temperatures (<500°C) produced more pronounced plateaus of the Tc vs. oxygen curve, with Tc ~ 91 K in the range x > 6.9, Tc ~ 60 K for 6.5 > x > 6.4 and loss of superconductivity at x ~ 6.45. Another reason for using quenching has been to avoid the appearance, during cooling, of a mixture of phases with various defects and oxygen ordering reactions. These were shown to exist by electron microscopy (EM) studies (e.g., Van Tendeloo et al. 1987, Van Tendeloo and Amelinckx 1990, Cava et al. 1990) and proposed by various theoretical models (e.g., de Fontaine et al. 1990a,b, Semenovskaya and Khachaturyan 1993). As the mobility of oxygen was considered to freeze at lower temperatures, quenching was used to produce "pure" phases. As we will see, the mobility of oxygen is appreciable even at low temperatures, so there is no need for quenching. Slow cooling leads as near as possible to the thermodynamic equilibrium state (sect. 3.2.2). A different method, although still using quenching, is Zr-gettering, developed at Bell Labs by the late J. Remeika (Chaillout and Remeika 1985). This method has been used extensively by Cava and colleagues, and combined with moderate quenching it gave as we will discuss later - good results.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30~

13

3.1.2. Synthesis near thermodynamic equilibrium A small group of researchers preferred synthesis near thermodynamic equilibrium. Jacobson et al. (1989) applied very early the equilibration of mixtures of 123-O6 and 123-O7 in a closed system. A similar method was used by Radaelli et al. (1992). Conder et al. (1994a,b), Krfiger et al. (1994, 1997), Kaldis (1997) and Kaldis et al. (1997b) used equilibration via the vapor phase of locally separated 123-Ox and Y metal for their nearequilibrium samples (table 2). Annealing in a reducing atmosphere was also used (e.g., Farneth et al. 1988, Monod et al. 1987); near equilibration in an open system (O2-Ar mixtures) was applied, e.g., by Rusiecki et al. (1990) for x/> 6.45. Electrochemical titration in a solid-state ionic cell was applied by Beyers et al. (1987) and Ahn et al. (1990). Its reversibility is an appreciable advantage. Table 2 Synthesis conditions for equilibrium samples Expelling carbonate (BAO method)

Ba-metal oxidation (DO Method)

Carbonate as Ba source. M i n i m u m carbonate content (CAR Method)

Carbonate as Ba source

Rusiecki et a1.(1990), Kaldis et al. (1997a)

Conder et al. (1994a), Kaldis et al. (1997a,b)

Krfiger et al. (1994, 1997), Kaldis et al. (1997a,b)

Jacobson et al. (1989), Radaelli et al. (1992)

Oxidation of Ba at low oxygen partial pressure; Solid-state reaction with CuO and Y203 in flowing oxygen; Further like CAR method

Solid-state reaction of Ba2CO 3 (4N) in oxygen with CuO and Y203; 6 annealings with T increasing 860 ~ 935°C, each time cooling at RT and grinding; Annealing at 935°C for 75 h, 50h cooling at RT 30°C/h, grinding; Annealing at 935°C for 80 h in oxygen, then cooling 10°C/h; x = 6.990; Diamagnetic transition AT(10%-50%) = 3 K (cf. fig. 8, for 6.912)

Solid-state reaction of Ba2CO 3 (4N) in oxygen with CuO (5N) and Er203 (5N); 900°C 12 h, in flowing oxygen, slow cooling, several times grinding and reheating up to 970°C; Final grinding; Annealing at 500°C; slowly cooled oxygen-rich material x ~ 6.9; Diamagnetic transition A T ( 1 0 % - 9 0 % ) = 15 K Oxygen-poor 123-0~ x ~ 6.1 with vacuum annealing

With Y-turnings (or Cu, or 123-O65 ) like the CAR samples

With Y-turnings (or Cu, or 123-O6.5) in separate A1203 crucibles placed in evacuated ampoules; 600°C for 15h; cooling 10°C/h

Mixtures x = 6 . 9 and x=6.1 in different proportions, sealed in evacuated ampoules; Annealing 14 days, slowly cooled in steps of 50°C, down to 50°C, in two months

Master batch

Ba2CO 3 ~ BaO 950°C, 40 h; Solid-state reaction with CuO and Y203 in flowing oxygen; 3 X 950°C, 20h; interm, grindings

Reduction

With gas mixtures

(At + 02)

14

E. KALDIS

Measurements of the diffusion coefficient of oxygen by isotope exchange showed that at least down to 250°C thermodynamic equilibrium exists (Conder et al. 1994a,b, Conder and Krfiger t996, Krfiger et al. t994). Quenching in liquid nitrogen and annealing at temperatures as low as 0°C have shown changes of Tc due to the mobility of the oxygen atoms (Veal et al. 1990a, Claus et al. 1990, Jorgensen et al. 1990b). Recently, it was also found that the mobility of oxygen in another important cuprate, LazCuO4+y, remains down to 200K (Wells et al. 1997). Slow cooling down to room temperature was applied under a controlled atmosphere in order to produce samples as "near as possible" to the thermodynamic equilibrium. This cautious expression is necessary in view of the complex structure of the HTc cuprates which may hinder complete relaxation even in polycrystalline samples. These equilibrium samples show hitherto unknown properties which shed new light on the problem of the intrinsic inhomogeneities of the HTc cuprates (Rusiecki et al. 1990, Conder et al. 1994a,b, Kaldis 1997, Kaldis et al. 1997b) (sects. 5.3, 6, 7). In addition to the thermal history, the chemical synthesis reaction also contributes to the formation of defects. In particular, the carbonate content of the starting materials and the precursor reactions for its elimination are important (sect. 3.1.4). Thus, several methods of preparation can be derived for equilibrium samples. The general strategy is to synthesize samples as oxygen-rich and as impurity-poor as possible (sect. 3.1.2.1), and then decrease their oxygen content by equilibration with a reducing agent having an affinity to oxygen comparable to that of 123-Ox. To avoid any contaminating components, yttrium, copper metal or oxygen-poor 123-Ox have been used as reducing agents (sect. 3.1.2.2).

3.1.2.1. Ooerdopedsamples, x > 6.95. Three different methods may be used to keep very low the carbonate content of equilibrium samples. High-purity oxides, Y203, CuO and BaCO3, and high-purity Y and Cu metals, all 4N (99.99%), have been used as precursors. Only Ba metal was 3N, the main impurity being Sr. (A) Expelling carbonate (BAO samples). BaCO3 is decomposed in an oxygen atmosphere to BaO, by firing for a long time (40 h) at high temperature (950°C) to expel the carbonates. To avoid recapture of carbonate the oxide is taken from the furnace when still hot, and then introduced in a glove box with an argon atmosphere (gettered with hot cerium turnings to react with traces of oxygen, carbonate, hydrocarbons and humidity). There, it is mixed by grinding with the other oxides and then reacted in a furnace with a flowing oxygen atmosphere at 950°C for 20 h. Three grindings and annealings are performed under similar conditions, the last annealing in ceramic pellet form. These oxygen-rich samples are then reduced to lower oxygen contents by controlled reduction with O2-Ar mixtures. Due to the very low oxygen partial pressure necessary, sometimes, at this early period the equilibration could not be completely reached in one step. That is why the scattering of lattice parameter data is higher than that of the CAR samples (sect. 3.2.2). Nevertheless, work with these samples showed for the first time anomalies of the lattice parameters, the existence of a maximum in the Tc vs. x curve (Rusiecki et al. 1990) and the dimpling transition

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN YBa2Cu3O4

15

with elastic neutron scattering (Conder et al. 1994a) (sect. 6). Judged from these results, it seems that this method is sensitive for metastable phases (sect. 3.2.2.4). (B) Ba source without carbonate, directly oxidized Ba metal (DO samples). Oxidation in a closed, evacuated system with controlled partial pressure of oxygen to avoid burning of barium, followed by a reaction with the other binary oxides. The experimental setup did not allow simultaneous processing of the whole batch, and therefore, some scattering of the data from these samples appears. Some of this scattering was found to be of an intrinsic nature, indicating pinning of the structure by phase transitions or latent structural changes in the "homogeneity" range of 123 (sect. 3.2.2.4). (C) Carbonate as Ba source, minimum carbonate content (CAR samples). Ahnost all the investigations in the literature have been performed with samples synthesized with BaCO3. With the exception of the work mentioned above - in BAO and DO samples - the non-linearity of the lattice parameters as a function of the oxygen doping (Rusiecki et al. 1990, Conder et al. 1994a, Kaldis et al. 1997b, KNger et al. 1997) (sect. 3.2.2, 5.4.2) was not discussed elsewhere in the literature. It was imperative, therefore, to investigate samples synthesized with a BaCO3 precursor under near-equilibrium conditions and compare them with those of samples from the above non-carbonate methods and the literature. Ba carbonate (99.99%) was reacted in an oxygen atmosphere with the other highpurity oxides in 6 annealing steps (each 35 h) interrupted by grinding in the glove box. The temperature of the first step was 860°C in order to avoid melting due to the formation of eutectics. This would form compact material and hinder the reaction kinetics. The subsequent steps were performed with increasing temperature, up to 935°C. This last annealing was repeated twice for 75 and 50 h and was followed by cooling to room temperature with a rate of 30°C/h. The final step was annealing for 80 h in oxygen and then slow cooling down to room temperature with 1-10°C/h. As shown with mass spectrometry combined with thermogravimetry (Macejewski et al. 1994) (sect. 3.1.4) the above procedure (Kffiger et al. 1997) leads to a minimum of carbonate content, due to the extended high temperature annealing. The phase purity was determined with microprobe and calibrated X-ray powder diffraction. Only traces of BaCuO2 were detected. The detection limit was estimated using powder mixtures of YBa2Cu306.8 with exactly known amounts of BaCuO2 in the range of 1-10 wt.%. Thus, it was possible to deduce by extrapolation a detection limit of 0.5 wt.%. The actual BaCuO2 content of the CAR samples was below this limit. The oxygen content was x -~ 6.990. A similar method, discussed in sect. 8, has been used for the synthesis of 11 Ca-doped and one O-overdoped 123 samples. 3.1.2.2. Optimally doped, and underdoped samples, x < 6.95. Reduction of these highoxygen-content DO and CAR samples to achieve various oxygen nonstoichiometries is performed by equilibrating with an exactly known weight of Y-metal turnings in sealed ampoules. Alternatively, Cu-metal turnings or 123-Ox with lower x value can be used. Sample and reducing agent are placed in separate small alumina crucibles in evacuated,

16

E. KALDIS i

0.0

•

i

i

a (x=6.912)

-0.2 E :

-0.5

-0.8 -1,0 80

I

l

84

i

I

88 T (K)

I

I

92

Fig. 8. DC normalized magnetization curve (SQUID, H = 10 Oe, field cooling) of x = 6.974 and x = 6.912 compositions. The high temperature resolution (0.1 K) allows a very high density of measurements. The solid lines represent a twoparameter fit of the Shoenberg model which perfectly describes the optimally doped sample x=6.912. For discussion see sect. 7.3. After Conder ct al. (1994b).

horizontal ampoules. The reduction is performed at 600°C for 15h, and the furnace is cooled with 10°C/h. Alumina does not react with the cuprates at these low temperatures. 100 samples covering the complete nonstoichiometric range 6.00 ~<x ~< 6.990 o f tetragohal and orthorhombic 123-Ox were synthesized with this method. An advantage o f the CAR samples seems to be their increased apparent homogeneity. Measurements o f their structural and physical properties show much less scattering. This is particularly evident in lattice parameters and micro-Raman measurements investigating with a microscope individual microcrystallites. This increased homogeneity is possibly due to traces o f carbonate functioning as flux (homogenizing agent). However, this homogeneity is not necessarily intrinsic (sect. 3.2.2). It tends to mask transitions and metastable phases. It is interesting to make a comparison o f the methods o f synthesis described here (sect. 3.1.2.1 and 3.1.2.2) with some well-equilibrated samples using a BaCO3 precursor in the literature (Jacobson et al. 1989, Radaelli et al. 1992). Two points are probably detrimental in their experiments: The expulsion o f the carbonate at higher temperature (970°C) which introduces traces o f decomposition products, and the equilibration o f a mixture o f 123-O6.1 and 123-O6.9 powders in sealed ampoules. As the authors observe themselves (Radaelli et al. 1992), this method leads to a slightly inhomogeneous product in the range o f the T --+ O transition. Thus, the first-order phasetransition line is crossed for different crystallites from opposite directions, causing c-axis discontinuities at different compositions. The same problem arises if, due to kinetic reasons, equilibrium is not attained at the same rate from both sides. Thus, the abovementioned local separation o f the reducing agent (Conder et al. 1994a, K N g e r et al. 1997, Kaldis 1997) has clear advantages. Polycrystalline, stoichiometric samples (x=6.912) prepared with the above methods have d.c. susceptibilities with sharp diamagnetic transitions, as shown in fig. 8 (Conder et al. 1994a) for the optimally doped material with ATc(10-50%)=2.8 K (el. sect. 7.3, figs. 86, 87).

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30 ~

17

3.1.3. High-accuracy determination of the oxygen content Stoichiometry control cannot be better than the accuracy of the determination of nonstoichiometry. In the literature the nonstoichiometry of 123-Ox has been determined with various methods: Iodometry. Standard iodometry has been used (Hume and Kolthoff 1944, Harris et al. 1987) with an error limit, but more sensitive is the use of bromide (Appelman et al. 1987). Often the achieved error limit is larger (Ax=+0.02). Details of the methods are given by Harris (1990). Thermogravimetry. More popular, because less time consuming, is thermogravimetry with error margin Ax ~> :E0.02 (see, e.g., ±0.04, Jorgensen et al. 1990a). Conder (2000) calculated the shifts of Tc (ATc/AX) corresponding to an error of Ax=0.01. Figure 9a reminds that, as expected, the error in Tc is not linear with x but oscillates between - 2 K (for x = 6.980) and +4 K (for x = 6.450)! Thermogravimetric reduction. One should be very cautious with thermogravimetric reduction in reducing atmosphere, usually hydrogen (Gallagher et al. 1987, Swaminathan et al. 1988). A certain advantage is the universality of the method, as it can be used for all cuprates, independently of the copper valence in the sample, as the final state is metallic copper. Also, it contributes to its popularity that no wet chemistry is involved. A plateau appears in the thermogravimetric curve near 650°C. Karppinen et al. (1996) proposed that this marks the stability range of Cu20. Conder (2000) found, however, from the weight loss that only a mixture of 2 B a O + 3 C u O + Y O O H seems to fit to this plateau. Unfortunately, the existence of YOOH has not yet been proved experimentally. In the cooling curve there is a plateau in the 750-550°C range which can be used for the determination of the endpoint (Graf 1991). At lower temperatures, the increase of weight is probably due to a reaction of BaO with H20 and/or C Q . An additional source of error could appear if the temperature difference of the weight measurement at the beginning 100.

a 80.

~. 6040. 200

4

g

3

~

2

? ~

1 o -2

6.4 ' 61.5 ' 6t.6 ' 617 ' 61.8 ' 6~.9 ' 7.0

Oxygen content

Fig. 9a. Lower part: shifts of T~ caused by an error of zXx=0.0] in the determination of the oxygen content. As the slope of the T~(x) curve is not constant the errors in T~ vat'y between - 2 and +4 K!. Upper part: Tc(x). After Conder

(2000).

18

E. KALDIS

(room temperature) and the end of the reaction (high temperature) is neglected. Also the existence of volatile impurities in the samples (CO2, H20) may add to the errors. Volumetric method. One order o f magnitude more accurate is a volumetric method (Conder et al. 1989). This method is based on an old principle of analytical chemistry and has been used in the past for the determination of the hydrogen in RE hydrides (Bischof et al. 1983, Conder and Kaldis 1989). The accuracy of the hydrogen determination is lower, because hydrogen is lost easily by diffusion through the seals (O-rings etc.) and the walls of the analytical apparatus. As this is not the case for oxygen, the extreme accuracy of Ax= 4-0.001 is achieved, for single-phase 123-Ox. Rusiecki et al. (1990) have calculated errors for the common second phase BaCuO2. For 5 mol%, which is clearly above the limit of detection by X-rays, the analytical error for an oxygen-poor sample near x = 6.53 would be Ax=+0.0006; and for an oxygen-rich sample near 6.94 Ax=+0.0081, i.e. still under the limit of detection of iodometry. As mentioned above, equilibrium samples may sometimes have up to 0.5 wt% BaCuO2! The application of this method to HTc superconductors is based on the oxidizing power of the (Cu-O) +p complex, shown by its reaction with acid, water and other oxidizable species. Thus, when superconductors containing holes, i.e. (Cu-O) ÷p, come in contact with water or acid, oxygen is evolved. The important point is that this oxygen is not the result of the oxidation of the solution but its origin is the solid superconductor, as shown first by Shafer et al. (1988). Assuming an intermediate hexahydrate this process is schematically written for 123-Ox as 123-Ox~[Cu(H20)6] 2+ P--+[Cu(H20)6]2+ + 0.25 02.

The quantity of oxygen evolved from the decomposition of water is a measure of the hole concentration of the 123-Ox sample (Shafer and Penney 1990). In the case of the dissolution of the sample in dilute nitric acid the hole concentration is quantitatively calculated after the formal (quite simplified) reaction (Conder et al. 1989) n Cu~o~.+ in H20---~¼n O2 T + nil+ + n Cu2+. It is necessary to recall, however, that according to early XPS (Nficker et al. 1987), EELS (Nficker et al. 1988), and XAS (Bianconi et al. 1987) investigations, the positive charge is generally considered not to be on the Cu but on the (Cu-O) + bond, and 80% on the oxygen p-orbitals (see e.g., Mfiller 1994). The great absolute accuracy o f this analytical method lies in the very accurate calibration. This is done by producing the necessary oxygen in the apparatus not by the decompo-

sition of a sample but by electrolysis ofH20. Coulometry is one of the most accurate physical measurements and therefore permits both high sensitivity and absolute accuracy. The method needs ~100mg for each determination, which is no problem for polycrystalline samples, but is unfortunately too large for most of the currently available single crystals. 2-3 determinations are made from each sample. Samples of the composition x ~<6.45 cannot be analyzed with this method because no excess holes exist when the superconductivity disappears. The method is applicable also to La cuprates (Conder 1999).

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30 x 0 10

....

1 , ....

2 , ....

3 ~ ....

4 , ....

CO inlet

9- "

11

8 _-

5 ,,,,

I

6 , ....

7 , ....

8 , ....

9 , ....

10 10

12

9

~

8

10

7-

19

Ill

A

7

.

6-.

1

'7

4

6

5-

5

4-

2

-

2a

5

6

,

0

.... 0

, ....

, ....

, ....

1

2

3

, .... 4

It , ....

, ....

, ....

, ....

, ....

5

6

7

8

9

0 10

Fig. 9b. Apparatus for the volumetric oxygen determination of 123, after Conder et al. (1989) (schematically): (1) Flowmeter for the pure CO2 stream. (2) Reaction vessel containing some diluted nitric acid, a piece of a stirring magnet, and the fragile glass capsule (2a) containing the 123 sample, which is broken with an external magnet. (3) Trap for KOH, filled with mercury, to avoid neutralization of the acid in the reaction vessel. (4) Absorption chamber filled with KOH to absorb the CO 2. (5) Niveau vessel joined with a long flexible tube to the absorption vessel, to change the niveau of KOH in (4). (6) Plexiglass rod with conical tip, operated vertically, to open and close the conical ground glass join of valve (7). (8) Water thermostat (T = 22.50±0.01°C) to keep the gas burette (10) and the oxygen volume at constant temperature. (9) Micrometer-driven syringe for the measurement of the volume of the evolved oxygen. (11,12) Valves (like 6) enabling the circulation of KOH and oxygen in the various chambers of the apparatus.

The high accuracy of this method permitted for the first time the study of oxygen doping of 123-Ox in very small intervals. This is conditio sine qua non for the discovery of latent phase transitions and lattice changes (Rusiecki et al. 1990, Conder et al. 1994a, Kaldis 1997, Kaldis et al. 1997b) (sect. 6). Figure 9b schematically shows the apparatus for oxygen determination. The 4N8(!) pure CO2 gas (constant flow rate 40 cm3/min) carries the oxygen evolved by the reaction of diluted HNO3 acid (1HNO3 : 9H20) with Y-123-Ox (in reaction vessel 2) into the absorption chamber (4). There the CO2 is absorbed by the KOH solution and the oxygen to be measured gathers at the entrance of valve 7, and is then measured in 9 (or 10). At the beginning of the experiment, after purging the system for 20 min. the rest-oxygen value of the apparatus is determined, at least two times. If the gas flow system is clean and tight, the blank test value is in the range 0.006-0.015 cm 3 for the gas purity used. This value has to be subtracted from the oxygen volume measured after breaking the sample capsule. At the end of the experiment another blank test is measured. If the flushing of oxygen is complete the initial and the final blank test values have to be equal.

20

E. KALDIS

After the blank rest-oxygen value is found constant in at least two determinations giving equal values, the evacuated glass capsule containing the sample is broken in the reaction vessel 2 and the carrier gas transfers the oxygen into the absorption chamber 4. When all the gas is transferred, valve 7 is opened and the gas is measured either with the thermostatized (-4-0.01°C) gas burette with measuring accuracy ~-0.1 mbar, or, preferably, in a micrometer-driven syringe with measuring accuracy 4-0.001 ml. Depending on the sample weight and composition an oxygen volume of 1.5-3.0 cm 3 is measured. As mentioned above, the high accuracy of the method is due to the calibration with electrolysis of water. This is done by changing the reaction vessel 2 with the oxygen electrode of the electrolysis of water (Conder et al. 1989). The current is coulometrically controlled with very high accuracy and, therefore, the amount of oxygen introduced into the apparatus also. For further information the author can be contacted. Coulometrie titration. This method was used by Ahn et al. (1988, 1990) for the determination of phase equilibria, and by Tetenbaum et al. (1989) for the determination of a possible miscibility gap. A drawback is that the absolute accuracy depends on the oxygen content of the starting material. This was determined by iodometry and therefore the accuracy was only Ax= 4-0.02. The near equilibrium conditions of the synthesis of their samples allowed the determination by EM of the various superstructures (sect. 5.1.1). 3.1.4. Mass-spectrometric determination of the carbonate content The incorporation of carbonate in the cuprates was investigated from the early stage of the HTc era and was found to decrease the superconducting properties and particularly Tc (e.g., Gallagher et al. 1988, Lindemer et al. 1990, Greaves and Slater 1991, Wang et al. 1996). These investigations later triggered the synthesis of a series of copper oxycarbonates, some of them with appreciable Tc's (e.g., Domanges et al. 1993, Raveau et al. 1993). The CO2 was measured with various methods e.g., by heating up to 1100°C and measuring with a Leco carbon analyzer (Lindemer et al. 1990), absorbing in Ba(OH)z (Karen and Kjekshus 1991) or using DTA and thermogravimetric methods (Gallagher et al. 1988). Structurally, the number of carbons involved was estimated in 123 from the number of C - O short bonds (Greaves and Slater 1991). Thermal desorption combined with mass spectrometry were performed up to 900°C (Keller et al. 1987) and the desorbed gas composition was found to be 0.59 wt.% CO2 and 0.28 wt.% CO. In a later investigation up to 820°C (Wada et al. 1992) no quantitative determination of the carbon-containing species was made. Using a combination of a thermobalance and a mass spectrometer DO and CAR samples (sect. 3.1.2.1) were investigated (Macejewski et al. 1994). A highlight of these investigations is that they led to low carbonate contents of the order of 2000 wt. ppm or less. Part of this carbonate is due to contamination from the atmosphere. As an example, a sample synthesized with BaCO3 precursor and annealed for 12 h at 900°C in oxygen atmosphere is shown in fig. 10 (curve a). As a result of this rapid synthesis, it contained ~5-10% B a C u Q and Y2BaCuO5 as shown by XRD. The thermogravimetric determination gave a total of 5800wt. ppm CO2 (0.58 wt%). Following a procedure

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30x

21

960

A

E

1040°C

E

E

8;0

oo' lO'00 Temperature [OC1

600

t a.

.<

i

•

=

•

i

.

=

.

11'00

°/

2oo

Fig. 10. Simultaneous thermal and massspectrometric analysis of a CAR sample (Ba source: carbonate, sect. 3.1.2.1) heated in oxygen stream: (a) evolution of CO2 from a rapidly synthesized sample (CH- 123, 5800wt.ppm) and the same sample after CAR procedure annealings (LPC 35, 1050wt.ppm); (b) DTA and TG curves. The desorption of CO2 takes place at three temperatures, indicating incorporation at three different sites of the lattice. After Macejewski et al. (1994).

,

t

,'f

b

E

855

_Z"

#= ~5 ,= 3

T

V ,0.oc

E 600

I

l

I

700

8~

9~

l

Temperature

l

971 l

;

1~0

[°C]

11 0

1200

Fig. 11. As fig. 10, but for a DO sample (sect. 3.1.2.1), Ba source: metal not carbonate: (a) I-A08-1 sample stored 250h in CO: atmosphere at room temperature, and I-A08-2 at 120°C.Note the shift of desorption temperatures and intensities (concentrations) indicating changes of the CO2 bonding depending on adsorption temperatures. Sample I-A08-2 contained nearly twice as much CO2. Experiment in oxygen atmosphere. (b) DTA signal of I-A08-2. After Macejewski et al. (1994).

similar to that o f the C A R samples, the same sample was then annealed exhaustively 2 x 2 4 h (in air), 2 × 4 8 h (air) and 5 h (air), each time with intermediate grindings (total o f six). As a result, only a small peak r e m a i n e d at 1047°C, with a total C Q content o f 1 0 5 0 w t . p p m (fig. 10, curve b). This is a temperature at which the c o m p o u n d decomposes (>970°C) and, therefore, this residual CO2 content can be r e m o v e d only i f the C A R sample is destroyed. The Tc's for figs. 10a and 10b were the same, b u t the transition width o f the a sample was four times that o f b. Figure 11 shows the behavior o f a D O sample. The same three peaks o f CO2 evolution were found (cf. fig. 10a), indicating the existence o f three different sites o f incorporation in the 123 lattice. However, whereas

22

E. KALDIS

the total residual amount at 1043°C was 2350 ppm (similar to the above), more than 50% evolved at lower temperatures. The reason for the existence of a similar residual CO2 in the DO and CAR samples (sect. 3.1.2.1) must be attributed to the closed system in which the oxidation of the metals and all annealings of DO took place. The desorbed gas was absorbed again during cooling. Annealing in an open system with flowing oxygen would decrease the carbonate. The origin of CO2 in the DO samples is very probably in the starting materials and the exposure to air. Experiments in argon atmosphere showed that the evolution from the CAR samples is coupled with evolution of O and therefore decomposition. This supports the strong bonding of CO2 in CAR and the conclusion that it would be possible to expel most of the carbonate from the DO samples without decomposition, but not from CAR samples. As we will discuss later (sect. 3.2.2.5), some characteristic structural anomalies of the lattice constants do not appear or are different for CAR samples. We note that almost all samples used in the literature have been synthesized with a BaCO3 precursor. 3.2. Changes of the lattice parameters and bond lengths with the oxygen

nonstoichiometry: the structural T---+0 transformation 3.2.1. Nonequilibrium samples, mainly underdoped regime 3.2.1.1. Lattice parameters of non-equilibrium samples. Figures 12a-c show the dependence of the lattice parameters, on the oxygen nonstoichiometry from the data of Cava et al. (1990) and Jorgensen et al. (1990a). We note in the data of Cava et al. (1990) the appreciable stepwise contraction of the c-axis by ~0.6% in the x ~ 6.35-6.45 range where the T ~ O structural transformation takes place. Similar steps appear also in the aand b-axis. This effect appears both at RT X-ray investigations (Cava et al. 1988a,b) and LT neutron diffraction refinements (Cava et al. 1990). Electrical transport and magnetic measurements showed that the onset of superconductivity (insulator-metal transition) for these samples takes place at x = 6.45. In sect. 3.2.2 we will see that equilibrium samples"

do show also abrupt steps of the lattice parameters. The abrupt change of the lattice parameters has been contested by many other investigators, the reason being, like in so many HToS issues, the different thermal and chemical histories of the samples. Jorgensen et al. (1990a) found in their RT neutron diffraction investigation a change of the slope of the c-axis but not an abrupt step, for samples quenched from higher temperatures than those of Cava et al. (1990), and without gettering (fig. 12c). This has been confirmed also by later investigations with samples equilibrated at lower temperatures (e.g., Radaelli et al. 1992). The step appears also in the apical bond (Cu2-O1) (fig. 14). Cava et al. (1990) concluded from this abrupt structural change that the superconductivity is triggered by the T---+ O transition and, therefore, it appears at the onset of the orthorhombic phase. Their paper presents the first systematic interdisciplinary investigation of the structural parameters and magnetic properties as a function of the nonstoichiometry, supporting the charge-transfer model from the chains to the planes which is now generally accepted. From the measurements of fig. 12 and many others at the time, the general impression remained that in all other ranges o f x the lattice parameters change linearly. Investigations

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN YBa2Cu30x i

,

i

,

i

.

i

J

. i

. i . i

23

, J , i

(a) ° / ° - - ¢

3,863 , 8 5 -

' ~

!

3 , 8 4 -

E

"o~.

Jorgensen et al. o

Cava et al. 1990

3,83-

3,82: 3,81

i . i , i • i • i 6,0 6,1 6,2 6,3 6,4

i , i , i . i , i , t 6,5 6,6 6,7 6,8 6,9 7,0

Oxygen content x '

(b)

3,89

Jorgensen etal.

'

I

123-0 x

/f J\

3,88 "~

'

3,87'

E ~.

o/O--c

3,86-

I

.6 3,85 610

6:2

6,4

6,6

7,0

6,8

Oxygen content x 6,0

6,2 .

11,84:11,82-

.

.

6.4

,

,

6,6

:,

,

.

6,8

,

.

.

.

7,0

,

.

.

.

(c)

11,84 11,82

11,80-

11,80

11,78-

[ii

11,76 i 6=:

11,74

orgensen et al. 1990 -o~

11,78 11,76 11,74

11,72 J~" 11,70-

- -

11,68~ .

6,0

.

.

,

6,2

,

,

1 1 , 7 0

~

::

123-0x

11,66 ~

~

i .

,

11,68

C a v a e t a l . 1990 ~ . ,

,

6,4

.

,

,

,

r

,

.

6,6

Oxygen content x

,

6,8

.

11,66 ,

,

7,0

Fig. 12. Dependence of the lattice parameters on oxygen nonstoichiometry:(a) a-parameter; (b) b-parameter; (c) c-parameter. NPD after Cava et al. (1990) (solid squares), at T=5K (compare table 3a); Jorgensen et al. (1990b) (open circles), NPD at RT. Note the abrupt change at the phase transition (x = 6.35-6.45) of the Zr-gettered samples of Cava et al. The vertical shift of the two groups of data is due to the differenttemperatures. The verticallines show the boundariesof changes of the lattice parameters. The onset of superconductivityfor the Zr-gettered samples is x = 6.45, that for the quenched samples is x ,-~6.35.

with a much larger number of equilibrium samples and more accurately determined oxygen content show several deviations from linearity for the a-, b-, and c-parameters with the corresponding changes of orthorhombicity and unit cell volume near the overdoped phase (sect. 3.2.2.3 and 3.2.2.4). The dependence of the structural parameters on the oxygen content of the gettered samples as refined by neutron diffraction in the centrosymmetric P m m m space group for the orthorhombic and in the P4/mmm for the tetragonal phase at 5 K (Cava et al. 1990) are given in table 3a. The experiments were not aiming at the highest possible resolution but combined good resolution (D2B instrument in ILL, Grenoble) with the investigation

24

E. K A L D I S

~1~ •

M M ~

~~ d

d

~ d

d

d

~ d

d

~ ~

d

d

~ d

d

~ ~

d

d

M

(D

P.

tt~ i~ " I

,.6

Z

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30 x 2.2

•

,

.

,

.

,

,

25

|

2.0

n(O(4))

1.8

o

7

7

0.. 0.8

0

0(1)) 0.6

CD 0.4

0.2 0.0

0.0

0.2

0.4

0.6

8

0.8

1.0

Fig. 13. Site occupancies for oxygen in the 123-O, structure as a function of nonstoichiometry from RT neutron diffraction. 6 ~ 7 - x is based on the measured weight loss. Note the use of O1 for the chain sites and 04 for the apical sites. A smooth T O transition takes place, indicating a second-order transition. This is in contrast to later experiments (Radaelli et al. (1992); cf. fig. 34, sect. 3.3). After Jorgensen et al. (1990a).

of 10 samples in the range x = 6.00-6.95. The highest error source is the reproducibility of the position of the samPoles in the beam to achieve the same wavelength. The estimated error limits are -t-0.0003 A and are not included in the values for the lattice parameters. The calculated standard error for the oxygen occupancies is 3%. All sites were refined and with the exception of the chain oxygen 0 4 found fully occupied. Jorgensen et al. (1990a) found for their samples (quenched from higher temperatures in liquid nitrogen) an up to 5% occupancy of the antichain position 05 and even some vacancies for the apical oxygen (fig. 13; note that in this figure 0 4 is used for the apical oxygen and O1 for chain oxygen). Later work with better equilibrated samples (Radaelli et al. 1992, B6ttger et al. 1996) confirmed the occupancy of 05. Single-crystal work by von Zimmermaun et al. (1999) did not show 05 site occupation (sect. 5.1.1). We have, therefore, to conclude that the 05 occupancy depends sensitively on the synthesis conditions. 3.2.1.2. Cu-O bonds. Bond lengths at 5 K (Cava et al. 1990) are given in table 3b. As mentioned above, the change of the O-nonstoichiometry strongly affects the apical bond Cu2-O1 (Cava et al. 1988a,b, 1990). Figure 14 shows this trend and fig. 15 shows the changes of the Cu2-O2 and Cu2-O3 bonds in the superconducting planes which are appreciably smaller (Hewat 1994). The 0.6% contraction step in the c-axis (fig. 12c) translates into a 3.4% contraction of the apical bond. A further 3% contraction takes place in the superconducting region (x > 6.4) (fig. 14), with increasing Tc (fig. 2). The same bond after Jorgensen et al. (1990a) is also shown in fig. 14. The influence of different synthesis conditions is again clearly seen. Although the trend of change of the bond with oxygen content is the same, differences in the slopes of the curves exist, particularly in the region of the tetragonal --+ orthorhombic transition at x ~ 6.4. The synthesis conditions of Cava et al. (1987a,b) - Zr getter, lower temperatures - and Jorgensen et al. (1990a) - quenched in liquid nitrogen from higher temperatures - are summarized in table 4 for comparison. Further discussion about the differences between the two curves in

26

E. KALDIS

¢;

,,6

;>.,

,6

v

×

×

I

]

,~ ~ 0 0

N

:x: ×

~ 0

r

~" 0 ~ I

I

×

~ 0

I

,:>-,

x

0"~ J

~7

~

~,~ I

I

u

0

r

~

I

0

I

~

"~" " " 0 0"~ I

I

r

;~ ~ I

I

~ i

o o o d

:~ ~ I

I

~

J

o

<

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30 x

6,0 ,

,

'~~ ...... j2,~o

6,2

6,4

2,40]

6,6

6,8

7,0

I ~+\oJ~aL1990]

6,0

27

6,2

6,4

6,6

6,8

j2,40

7,0

Oxygen content x

2,05-

Fig. 14. Dependence of the apical bond Cu2-Ol on the oxygen nonstoichiometry. Note the abrupt change of the Zr-gettered samples (solid squares) at the T--, O transition. NPD data, after Cava et al. (1990) and Hewat (1994) at T=5K, and Jorgensen et al. (1990a) (open circles) at RT.

--= Cu2.02] --,-- Cu2.03

2,00-

.<:
~¢,. 1,95-

1,900 an

1,85 -

Cavaet aL 1990 610'612'614'616'618'710 Oxygen content x

Fig. 15. Cu2-O3 and Cu2-O2 bond lengths as a function of oxygen nonstoiehiometry. After Cava et al. (1990) and Hewat (1994).

Table 4 Comparison of the synthesis conditions of quenched samples Cava et al. (1987a,b, 1990)

Jorgensen et al. (1990a)

Master batch, powder/pellets Fired in air 900 and 950°C 3 days Intermediate grindings. Fired in flowing 02, 16h at 1000°C; Annealing in flowing 02, 12h at 500°C Stated composition: x : 7.00? Impurities?

Master batch, powder/pellets Fired in aft 960°C 1 day Intermediate grinding (200mesh). Pellets 960°C 1 day in flowing 02; Annealing 670°C 1 day/cooling 50°C/h; Composition (iodometric) x - 6.93 -t=0.01 Impurities: few % green phase

Reduction of oxygen content 2 days 360-520°C with Zr foil" in sealed evacuated quartz ampoules;

Reduction 2~4 days 520:t-1°C O2-Ar controlled atmosphere at given Po2 (1-2× 10-4 atm);

Quenched to RT Oxygen determination Reduction in N2-15% H2; Stated accuracy Ax = 4-0.02

Quenched to liquid nitrogen Oxygen determination Iodometry: Ax = ±0.01; Thermogravimetry Ax = 5_0.02

a Cava et al. (1990). 4days at 440°C with Zr foil; variation of stoichiometry with the amount of Zr.

28

E. KALDIS i

i

!

t.9

i

0.5

t.8

0.4

o,,<

io.3 0

Z W J

<

2.0

Z 0 m

02 ~ , ~ , ~ , ~ C u

Cu t - 0 4

2-02,03 PLANE t.9

0.2

o.t 7

t.8 x in BozYCu30 ~

G

Fig. 16. Dependence of the puckering (dimpling) of the Cu2-O2,O3 plane and the Ba-O1 plane on oxygen nonstoichiometry. The separation is shown between the Ba and O1 layers, and between Cu2 and the average 02,03 layer along the c-axis. We note that the relative change at the step in the Cu2-O2,O3 dimpling is 2.3% whereas the bond lengths of Cu2-O2 and Cu2-O3 (fig. 15) change very little. Also, the Ba-O1 dimplingat x ~ 6.4 shows a change by 24% whereas the Ba-O1 bond lengthchanges only by 1% (fig. 18). After Cava et al. (1990).

1

7

I

I

x in Bo.YCu.O I _ .,-

!

G

Fig. 17. Dependence of the bond lengths between Cul and the apicaloxygenO1 and Cul and the chain oxygen 04 on oxygen nonstoichiometry.Whereas the effect is very pronotmcedin the first case (30%), it is very small in the second case. After Cava et al. (1990).

figs. 12 and 14 is given in the context of the T --+ O transition in sect. 3.3. The controversy about the existence of the abrupt step in the c-axis did not only come from the results of Jorgensen et al. (1990a) and other groups, but also from inconsistencies among the various Zr-gettered samples. Thus, Cava et al. (1988a,b) did not find this step and the onset o f superconductivity at the T - 4 0 transition, but inside the orthorhombic symmetry regime. Several other groups have also found the onset of superconductivity inside the orthorhombic phase (Maletta et al. 1989, Radaelli et al. 1992). As we will see, these findings seem to be confirmed by the equilibrium samples (sect. 3.2.2.3). The explanation given by Cava et al. (1990) is that this effect depends on the annealing temperature. Figure 16 shows the increase with oxygen doping of the dimpling C u 2 - O 2 / O 3 of the planes (i.e. spacing from Cu2 to the average 0 2 / 0 3 plane). The absolute change o f the

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBazCu30 x

29

mean value of the Cu-O2 and Cu-O3 bonds (fig. 15) is 0.055A and the relative change is significant: 2.3%. The dependence of dimpling on the oxygen content is discussed in sects. 5.3, 6 and subsequent sections for equilibrium samples. The changes of the Cul-O1 (apex) and Cul-O4 (chain) bonds are shown in fig. 17. The increase is 2.63% for the former, but only 0.8% for the latter. Summing up, appreciable changes of bond lengths take place between the regime of fourfold coordination (x> 6.35) and twofold coordination (x < 6.35) where the bond length remains constant (Cava et al. 1990).

3.2.1.3. B a - O and Y-O bonds. Figure 18 shows the changes of the Ba bonds to the apical oxygen O1 and the chain oxygen 04. The coordination changes stepwise again. The effect is much more pronounced in the Ba-O1 layer. The dimpling is shown in fig. 16. The bond lengths of the Ba-atoms to the oxygens of the planes 02 and 03 are shown in fig. 19. Like in the case of Cu2 (fig. 15) the orthorhombicity splits these values whereas in the tetragonal regime they overlap. Analogous is the situation for Y.

2.9

"

,

,

,

,

3.0

i

I

i

I

Ba - 0 2 Bo-Ol 2.9

2.8

I

-r t,Z

2.7

Z

,

,.~ 3 . 0

2.8

e~ 2 . 5

Z

2.9

2.8

Z

_/

o

f

o m

Y -02

2.4

I,

~

~

I

•

~;

Y - 03

8o-04

I

2.3

I

6 x in BozYCu30 ~

Fig. 18. Dependence of the Ba-O bond lengths on oxygen nonstoichiometry. The T--+ O transition is much more pronounced for the oxygen of the chains. After Cava et al. (1990).

I

t

I

7

6 x

In

BazYCu30

~

Fig. 19. Dependence of the bond lengths between Ba and Y and the plane oxygens 02 and 03 on oxygen nonstoichiometry. The splitting in the orthorhombic range is clearly seen. After Cava et al. (1990).

30

E. KALDIS

3.2.1.4. Application o f the Bond Valence Sum (BVS) method to 123-0x - I. The Bond Valence Sum (BVS) method is a very sensitive probe because it reacts to any change of the bond lengths of a solid. Due to this reason it is not very specific and the values it gives should be considered as good estimates of the valence and strain effects acting on a given atom. However, its great value is as a warning that non-ideal conditions are existing in the structure. Using a nonstoichiometric series consisting of many samples one can then find important trends between valence and other properties. The idea that the strength of the bond between a given anion-cation pair in a structure can give information about the valence of the ions involved originates from Panling (at a time where only very few crystal structures had been determined!) and is based on the idea that higher oxidation states lead to shorter bonds. Pauling (1929) defined a bond strength s for a cation-anion bond: s =z/N,

with z the formal valence of the cation, and N the coordination number of the nearest anion neighbors (assumed to be equivalent). Summation over the bond strengths of the cations surrounding an anion would give the anion valence. Much later, after a very large number of crystal structures had become available, it was shown that for most inorganic compounds the sum o f the bond valences around any atom lies within 0.1 of its formal oxidation state (Brown and Shannon 1973). The historical development of the method and the contributions of Zachariasen (1931) and Brown and Andermatt (1985) are briefly discussed by Marezio and Chaillout (1994) and by Hewat (1994). Brown (1989) presented a complete study of the oxidation states and the internal stresses in 123-Ox (x = 6.0-7.0) using the BVS method. He pointed out that because the structure of 123-Ox is highly constrained there are some important difficulties in applying this method. The bond valence s~j between a cation i and an anion j which are at a distance R/j, is given according to Brown and Andermatt (1985) by sii = exp(R0 - Rii/B ), where R0 and B are constants with tabulated values. B has a typical value of 0.37 A, and the R0 value was taken in 123-Ox to be 1.60A for Cu ~+, 1.679A for Cu 24, and 1.73 A for Cu 3+ (Brown 1989). These values arise as averages from a series of structure determinations of well-known compounds with corresponding copper valences, using the valence equation (1). From the Inorganic Crystal Structure Database Brown and Andermatt (1985) have determined average R0 values for 750 ion pairs. Summing over all the copper-oxygen bonds 0" in the copper coordination sphere, we can determine the valence of an ion (as a stun of bond valences) by solving the equation

(R0-R/:)

Vj=Zsg=~exp\ 0.37 J (1) J J With this equation Brown (1989) analyzed 13 structure determinations of 123-Ox with different x values. Initial calculations were performed for all copper as Cu 2+. When

OXYGEN

6,6 ,

NONSTOICHIOMETRY

6,4

6; I

66

AND

6,6

b

t~ X ILl

•.

.

/

•

•

alton

EFFECTS

1N YBa2Cu30

x

31

7,0

0,4- ~ ,

O,2-

LATTICE

0,4

.0,2

o,o-

"0,0

=. .-0,2

-0,2

I=

,~

989

. -0,4

-0,4 P

• -0,6

-0,6

6,0

,

i

6,2

'

i

6,4

'

u

6,6

~8

Oxygen C o n t e n t x

7,0

Fig. 20. Difference between the bond valence sums of the investigatednonstoichiometric 123-O~ samples and the oxidation states (per Ba2Cu3 formula unit) of Ba (circles, solid line) and Cu (squares, dotted line)• At x=7.0 the Ba is too small for its cage and at x=6.0 it is too big. Lattice distortions will appear at both these values (see text)• The diamonds show model calculations. After Brown (1989).

deviations appeared, mixtures o f the various copper valences were used. With increasing oxygen content the valence sums o f Y, 0 2 , and 0 3 remain more or less constant (at 3.0 and 2.0 respectively) and that of 04 increases (also of O1 and Ba). However, although the valence sums o f C u l and Cu2 increase with increasing oxygen content, their ratio to the expected value decreases due to the strains. A n important aspect o f this investigation concerns the internal stresses o f the 123-Ox structure and their change with oxygen nonstoichiometry. Undistorted cubic perovskites are very rare. Stresses appear in the following two cases: (1) I f the large atom A o f perovskite (Ba) is too large for its cubic cage BO3 (Cu3Ox) then the cage is under tension and Ba under compression. (2) I f Ba is too small for the Cu3Ox cage the opposite happens. Brown (1978) has shown that i f the difference between the valence sum determined from the crystallographic data and the ideal valence value is more than 0.2, the cubic perovskite structure is distorted by one o f the following two mechanisms: (i) W h e n A is too large for its cage the smaller B atom is shifted in an off-center position in its octahedron. This leads to aferroelectric phase, e.g., BaTiO3. (ii) W h e n A is too small for its cage the cage contracts by a rotation o f the octahedra leading to a ferroelastic phase. For the application o f BVS it is important to know that in the above cases the coordination sphere o f the cation under tension is distorted. This can lead to a higher valence sum without change of the average bond lengths (Brown and Shannon 1973). Thus, both

valence changes and lattice distortions affect the B VS. The differences between the valence sums based on the structural data and the oxidation states o f Ba and Cu - per formula unit Ba2Cu3 - as a function o f the oxygen nonstoichiometry (Brown 1989) are shown in fig. 20. The results indicate that oxygen-

32

E. KALDIS

I 2-01

" ,

• Cu1

t,5 :E D ~J

uz

t.0

~ 2"22 ~ 0 2 0

2.t7

Cu2

2.42

I 7

I

~

x In B o z Y C u s O z

l

Fig. 21. Bond valence sums of Cul and Cu2 as a function of oxygen nonstoichiometry.Note that the BVS of Cu2 follows very nearly the shape of the To(x) curve (fig. 2) and reflects the T ~ O transition and the optimally doped phase. The BVS of the chain Cul increases linearly due to the increase of the carriers concentration with doping. After Cava et al. (1990).

rich and oxygen-poor compositions are near to structural instability. Thus, for x ~ 7 the Cu3Ox cage is smaller than the Ba ion, leading to tensile stress. This tension is partly relieved by the orthorhombic distortion (elongation o f the C u l - O 4 bond = b-axis; increase o f the average C u 2 - O bond lengths by distorting the Cu2 environment). Additional relief comes also from the distribution o f the smaller Cu 3+ in the Cu2 and C u l sites, but the cages around the Cu ions remain too large. Brown (1989) concludes that further relief o f the tension could be expected from a transformation increasing the C u - O distances, e.g., a ferroelectric displacement o f the copper atoms from the centers o f their coordination polyhedra. For x ~ 6 the cage is too large for the Ba ion. The stress could be relieved if the cage would contract, e.g., by aferroelastic rotation o f the CuO4 planes. The BVS method has been extensively applied to estimate the carrier concentrations for a given To. For 123-Ox this is rather doubtful, due tot the existence o f the two different Cu sites, as will be discussed in sect. 5.3. Cava et al. (1988a,b, 1990) have calculated with the BVS method the formal valences o f the chain copper ( C u l ) and the plane copper (Cu2) as a function o f oxygen doping. Figure 21 shows that the C u l valence increases smoothly with oxygen content. Based on an ideal structure one would expect that C u l in 123-O6, with its twofold coordination to the apical oxygens O1 (fig. 6), would have a formal valence o f 1+ . The calculated 1.3 + value shows the structural strains which clearly exist in the 123-Ox structure, near the 06 stoichiometry, as Brown (1989) has shown. The - 2 9 % linear increase with doping in the superconducting range - 6 . 5 < x < ~6.9 ( - 4 5 %

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTS IN YBa2Cu30x

33

for ~6.0 < x < ~6.9) shows the strong increase of the charge associated with the chains. On the contrary, the change of the charge associated with the planes in the superconducting range is only ~2% and shows appreciable changes of slopes. In the range ~6.0 < x < ~6.3 no change of the Cu2 valence takes place. All the positive charges remain in the chains. In the range ~6.35 < x < ~6.45, the T --40 phase transition takes place and the superconductivity sets in. At this point the Cu2 valence shows an abrupt increase with doping. After filling two-thirds of the chains with oxygen, a smooth increase follows until a final plateau in the range ~6.84 < x < ~ 6 . 9 is reached in the optimally doped and overdoped regime. Very interesting is the similarity between the Tc vs. x curve and the formal Cu2 valence vs. x found by Cava et al. (1988a,b, 1990), supporting the charge-transfer from the reservoir block to the planes. The importance of the apical bond Cu2-O 1 for the charge transfer becomes apparent upon comparison of figs. 14 and 21. The contraction of the apical bond scales also very well with the increase of Tc as a function of doping. Based on their BVS results Cava et al. (1988a,b, 1990) could conclude that the decrease of Tc from 90 to 60 K (with decreasing O content) is due to the transfer of a negative charge of ~ 0.03e/Cu from the chains to the planes and that the superconductivity is suppressed if further ~ 0.05e/Cu is transferred to the plains. These and many other results led to the now accepted general picture of the HTc cuprates as consisting of copperoxygen superconducting planes sandwiched between charge reservoirs. Probably due to difficulties in synthesizing oxygen-rich samples at that time Cava et al. (1988a,b, 1990) have only one point over x=6.9. The same is true for Jorgensen et al. (1990a). Thus no conclusion can be drawn from these data about the phenomena in the optimally doped and overdoped regimes (sects. 5-7).

3.2.1.5. Mobility of oxygen, reordering at quenching. Underdoped (6.3 < x < 6.6) single crystals (<2 mg) quenched in liquid nitrogen from 520°C show at room temperature, stored in air, relaxation phenomena in both their structural and their superconducting properties. Veal et al. (1990a) have found an increase of Tc up to 15 K saturating upon aging after several days. It is accompanied by a decrease of the a-axis and an increase of the b-axis (increase of orthorhombicity), leading to a decrease of the oxygen content x of the T --+ O transition. The structural changes were investigated by X-ray diffraction. This effect is quite interesting because it shows that even at room temperature an appreciable mobility of the oxygen exists in the 123 lattice. This is in contrast to the assumption that at T < 500°C the mobility of oxygen is frozen, an argument for investigating quenched samples. Absorption of molecular oxygen and subsequent diffusion of atomic oxygen through the lattice seems to be too slow a process to account for this behavior. Several tests indicated that the changes were not due to absorption of additional oxygen from the air during aging (Veal et al. 1990a): (1) The final Tc values reached by the samples depend on the stoichiometries from which they were quenched. I f absorption of oxygen was taking place during aging then

34

E. KALDIS 25

. . . .

I

. . . .

I

. . . .

20

"-"

i

G

15

10

5

I

0

I

I

I

I

~

3000

,

~

,

i

6000

Time (rain)

,

i

,

,

9000

Fig. 22. T~.(o,~,) vs. annealing time at room temperature, t23-O6.41 quenched from 500°C to liquid nitrogen. Tc increases from 0 to 20K in ~150h due to local rearrangements of the oxygenin the chains. After Jorgensenet al. (1990b).

probably all samples would relax to the same oxygen content (under the assumption of comparable grain interface properties). (2) Absorption of oxygen at room temperature would be expected to cause broadening of the X-ray diffraction peaks. This was not the case. (3) Within the rather large error limit of a gravimetric determination (&~c=-t-0.04) no increases of the sample weight were detected. (4) In spite of the above, it could be argued that a shell effect of absorbed oxygen might result in the increase of Tc. This was shown not to be the case because a small single crystal crushed to powder still showed the same Tc. (5) Annealing an aged sample with x = 6.45 at 100°C once in flowing N2 and once in 02 and then quenching in liquid nitrogen, gave the same Tc's (however, both decreased by six degrees compared with the original). Based on this evidence Veal et al. (1990a) concluded that the increase of To and orthorhombicity observed were due n o t to additional oxygen absorbed during aging, but to rearrangement of the oxygen in the chains. They particularly proposed shortpath diffusion due to jumping from the anti-chain site 05 (statistically occupied in the tetragonal phase) to 0 4 during aging. This should result in changes of the hole concentration in the planes, increasing T~ and orthorhombicity in accordance with the literature (Cava et al. 1988b, Jorgensen et al. 1990a, Zaanen et al. 1988). To investigate the oxygen rearrangement mechanism Jorgensen et al. (1990b) used NPD, characterizing their polycrystalline sample (x=6.415 by iodometry) with dc magnetization as a function of the ageing time at room temperature. Starting with a To = 9 0 K (6.9 <x < 7.0) prepared after their previously published method (Jorgensen et al. 1990a), cf. table 4, they reduced the material in nitrogen containing 0.048% oxygen at 520°C for 112 h and then quenched in liquid nitrogen. The sample was zero-field cooled and then slowly warmed up in a SQUID magnetometer. Figure 22 shows the increasing T~ as a function of the aging time. The first measurement did not show superconductivity at all, but after 150 h the T~,onset had reached asymptotically 20 K, the same behavior as that

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS 1N YBa2Cu30x

35

of the single crystals above (Veal et al. 1990a). NPD showed that the unit-cell volume and the lattice parameters decrease and the orthorhombicity increases (a decreases more than b) with aging time (fig. 23) as Tc increases. A comparison of figs. 22 and 23 shows that in the initial stage these samples are orthorhombic but not superconducting, similar to some findings on slowly cooled samples (Maletta et al. 1989, Radaelli et al. 1992) and the equilibrium samples discussed in (sect. 3.3). We note the appreciable contraction of the unit cell volume in fig. 23 (0.1%), characteristic for the increased coupling with increasing Tc due to oxygen ordering. This results from the decrease of the c-axis (0.04%) and the a-axis (0.05%). In contrast, the b-axis contracts almost three times less (0.013%) indicating that an expanding mechanism may counteract. The timedependent increase of Tc (fig. 22) is given by the following equation, of a type often found for ionic and hopping relaxation phenomena: T~(t) = T~(oo) - [T~(cxD)- T~(O)]exp[-(t/T)l/2],

(2)

with t the annealing time and T a characteristic time constant with best-fit value 386 min. The large continuous decrease of the c-axis is a typical indication of the charge transfer in samples quenched from high temperatures (table 4, Jorgensen et al. 1990a), which do not show an abrupt step-like decrease like the Zr-gettered samples (Cava et al. 1988a,b, 1990). From the structure of the unit cell (fig. 5) we see that the c-axis is the sum of the C u - O bond lengths and the Cu2-Cu2 spacing 2 x ( C u l - O 1 + Cu2-O 1)+ (Cu2-Cu2). Therefore, the contraction of the c-axis must result from the contraction of some of these bonds or spacings due to charge transfer. Figure 24 shows the dependence of these bond lengths on the aging time after quenching. The main change is the contraction of the apical bond Cu2-O1 (Cu2-O4 in fig. 24) by 0.01 i which in spite of the small expansions of the C u l - O 1 and the Cu2-Cu2 leads to the contraction of the c-axis. The change of the oxygen site occupancies (O1, 04, 05) during aging was found to be insufficient for explaining the increase of T~ shown in fig. 22. Thus, it was proposed (Jorgensen et al. 1990b) that it is not the long-range oxygen ordering of the chains which leads to charge transfer and the increase of T~ after aging at room temperature, but the local oxygen ordering. It has been argued in the literature (see e.g., Hou et al. 1990, Zaanen et al. 1988) that the local ordering of the oxygens around Cul (04 and 05) influences its oxidation state via the coordination number and geometry. Roughly, these models assume that Cul can reach only the 2 + oxidation state. This is the case with three, four, five or six oxygen nearest neighbors, while 1+ is the oxidation state of Cul with two neighbors. A random distribution of the oxygens in the Cul plane leads to a maximum number of threefold coordinated Cu and therefore to a maximum of 2 + charges. However, when ordering leads to fourfold or higher coordination, the number of 2 + charges becomes a minimum. Then, the transfer of electrons from the planes to the chains reaches a maximum and the hole count in the planes increases. Thus, local ordering without a net change of occupancies may lead to an increase of the hole concentration in the planes and of To (Jorgensen et al. 1990b). This ordering is expected to cause an expansion of the b-axis, counteracting the general contraction of the unit cell (fig. 23). This is in agreement with the smaller contraction of the b-axis (fig. 23).

E. KALDIS

36 176.0

[

l

i

l

l

[

,

,

l

[

l

l

l

l

175.9

175.8

175.7 1t.779

l

,

l

l

l

l

I

,

i

,

L

i

r

,

l

,

,

I

,

,

l

,

i

,

,

b

,

,

~,~--~11.776 0

11.773 3.881

,

,

l

l

l

l

[

i

l

i

l

]

,

l

v¢-~ 3 . 8 8 0

3.879 3 . 8 5 0

, ~

,

I

,

,

[

,

,

I

,

,

I

,

,

i

[

,

i

i

i

,

I

,

i

~

,

,

A

3.849

3.848

'

~

;

300

0.032

,

i,,

I

I

600 900 Time (min) ~,,

1200

~,,

1500

L , ,

,,~ 0,031

v

0.030

0.029

I

300

I

I

I

I

I

I

600 900 T i m e (min)

I

I

[

1200

i

r

1500

Fig. 23. Relaxationof the unit cellvolume,latticeparameters,and orthorhombicstrainat roomtemperature, after quenchingfrom 500°C to liquid nitrogen. After Jorgensenet al. (1990b).

OXYGEN NONSTOICHIOMETRY AND LATT[CE EFFECTS 1N YBa2Cu30 x

37

3,33 .,< ~" 3.32

d& 3.31

3.30 2.42

'

'

i

,

,

i

,

'

I

,

I'-';

r

2.41

o EM

(,~ 2.40

2.39 1.84

,

,

1

i

l

i

l

I

i

=

I

i

'

'

1

,

,

I

,

i

,

~

i

,

i

~.1.83

o, 1.82

1.81

J

,

I

300

J

I

)

i

I

[

600 900 Time (rain)

,

[

i

1200

i

1500

Fig. 24. Relaxation of the bond lengths under the conditions of figs. 22 and 23. Note the contraction of the apical bond with time, here denoted as Cu(2) 0(4). This is the result of charge transfer without change of the oxygen occupancies, due to change of the local oxygen ordering. After Jorgensen et al. (1990b).

Soon after, Ceder et al. 1991a) performed a Monte Carlo sinurlation of the above Veal and Jorgensen experiments, with the same Ising model used in the ASYNNNI model to derive the phase diagram (sect. 5.2). The results indicated that in contrast to the above, O5-O1 ordering, which is faster because it needs only nearest-neighbor jumps, dominates the initial relaxation period (~90 min). This reduces the number of Cu with oxygen coordination 3, establishes the O-I structure (fully occupied chains) and increases the number of holes in the planes. This could explain why the samples quenched from the tetragonal structure are found to be orthorhombic. In a second stage, the much longer formation of the chain ordering dominates, leading to the O-II structure with every second chain empty (O-I to O-II). According to this sinmlation the above ageing experiments could observe only the second stage of the ordering. The transition from the tetragonal phase (statistical distribution of 05 and O1) to the O-I phase (no 05 sites occupied) occurred during quenching. Although this result is quite interesting, we note that, as discussed in sect. 5.2, the ASYNNNI model, on which this simulation is based, was not particularly accurate in predicting structural details. In conclusion, it can be said that the experiments of Veal et al. and Jorgensen et al. discussed above illustrate clearly the importance of the intra-chain local oxygen ordering.

38

E. KALDIS

Thus, chain length and inter-chain ordering in superstructures are important for charge transfer. Unfortunately, up to now a microscopic picture of exactly how charge transfer is controlled by these two parameters is lacking (cf. sects. 5.1, 5.2). 3.2.2. Equilibrium samples with high-resolution oxygen determination In sect. 3.1.2 we have discussed methods leading to near equilibrium samples with oxygen content up to x = 6.99. These methods, combined with an oxygen determination method 10 times more accurate than iodometry (Conder et al. 1989) and with a large number of investigated compositions, allowed to see more clearly the trends of structural, optical (Raman) and diamagnetic properties. They made possible a systematic study of the overdoped x > 6.95 range, and shed more light in the underdoped (6.95 ~>x ~>6.4) and insulating (6.40 ~>x ~> 6.0) ranges of 123-Ox. These investigations (1) Showed the existence of an up to then unknown dispIacive transformation between the optimally doped and the overdoped regime (sect. 6), and (2) Gave new results about: the transitions correlated with the onset of superconductivity (sect. 3.3.2); - a sequence o f steps in the in-phase Raman phonon frequency (sect. 5.5), - p h a s e separation in underdoped 123-Ox (sect. 5.5) - phase separation in oxygen-overdoped 123-Ox (sect. 7), and - phase separation in Ca-overdoped 123-O6.95 (sect. 8). -

3.2.2.1. Lattice parameters vs. oxygen content. Only a rough comparison can be made between the dependence of the lattice parameters on the oxygen nonstoichiometry for quenched samples (quenching from higher and lower temperatures, Zr-gettering etc.) and equilibrium samples. The reason is that the main systematic studies of the structure of 123-Ox as a function of the oxygen content using quenched samples (Jorgensen et al. 1990a, Cava et al. 1990) have investigated relatively few compositions in the range of x ,-~ 6.0-6.9. In this range, the high accuracy of the oxygen determination and the controlled near-equilibrium conditions allowed the synthesis and study of more than a hundred compositions with three different methods (table 2; Conder et al. 1994a, Kaldis 1997). In the case of CAR and BAO samples (sect. 3.1.2) more than forty equilibrium compositions each have been investigated in the above-mentioned x range. Thus, although there is good agreement in the range of the T - O transition, no comparison is possible in the overdoped region. We note that, as mentioned in sect. 3.1.3, high-resolution oxygen determination is not valid at x < 6.45, i.e. in the absence of Cu 3+ (formal valence). Samples with lower oxygen content were analyzed by iodometry and thermogravimetric analysis. 3.2.2.2. X-ray diffractometry, T = 30OK. The lattice parameters and some bond lengths of the three different kinds of samples discussed in sect. 3.1.2 have been measured with X-ray diffraction using either a Guinier camera with internal standard (Rusiecki et al. 1990, Rusiecki and Kaldis 1991, Conder et al. 1994a) or a STOE powder diffractometer

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTS1N YBa2Cu30~

39

(Krfiger et al. 1997) with Ge-monochromatized Cu Kal radiation (l = 1.54098). In the diffractometry, the parameters of the orthorhombic Pmmm space group were determined with more than 30 reflections. NIST silicon standard (a = 5.43088,~) was used as an internal standard for the determination of the lattice parameters (also with the film camera). Data were recorded in transmission geometry in the 2q range 5-120 °. Due to the low scattering power of oxygen for X-rays a refinement of the occupation of the oxygen sites was not performed. With the exception of the 0 4 site (chain oxygen), which was assumed to correspond to the value obtained by the volumetric oxygen analysis, all other oxygen site occupancies (including 05) had to be fixed. With the exception of Cul (chain copper), the metal occupancies were also fixed at 100%. The refinements converged at rather low disagreement factors between observed and measured intensities Re = 0.04-0.08. 3.2.2.3. Lattice parameters of samples with Ba carbonate precursor (CAR samples). Figures 25a-e show the abrupt change of the X-ray lattice parameters at the T---+ O transition. The stepwise change of the c-axis is 0.05A (0.5%). The step in the equilibrium samples is slightly smaller than in the Zr-gettered samples (0.6%, fig. 12c), but possibly more narrow. Less abrupt changes appear in the other lattice parameters, with the exception of the unit cell volume as we will discuss later. The difference from the quenched samples (Jorgensen et al. 1990a) is clearly shown in the example of the c-parameter in fig. 26a, whereas the difference in the in-plane axes is smaller (fig. 26b). We can, therefore, conclude that abrupt change of the lattice constant c at the T - O transition is characteristic for near-equilibrium samples, including the Zr-gettered synthesized under the conditions of table 4. DC susceptibility measurements (fig. 27a) show superconductivity at x ~ 6.396 for the equilibrium samples, but small amounts of the sample become superconducting at an earlier stage (fig. 27b; see also sect. 3.3). Another interpretation has been given by Tallon (1990) and Tallon et al. (1995) (see also review by Radaelli 1998) who proposed that the onset of superconductivity in 123-Ox appears at a certain value of the concentration of holes per Cu2 atom. As we will discuss in later sections (5.3, 8.4) this is still questionable, because good Yi-yCayBa2Cu306 samples seem not to be superconducting (Merz et al. 1998). Tallon et al., however, had superconducting samples. As already mentioned, although the transition appears very abrupt in the c-axis (fig. 25d) and in the unit cell volume (fig. 25e) it extends to a larger composition range 6.3-6.4(5) for the a-axis (fig. 25a) and 6.3-6.45 for the b-axis (fig. 25b). It could be argued that the transition starts at the a-b plane, the corresponding c-axis changes being too small to be seen in the 6.3-6.4 range. Synchrotron radiation measurements could clarify this point. The total relative change of the lattice parameters in the x = 6.0-6.99 range is large: -1.15% for the a-axis, +0.72% for the b-axis, -1.64% for the c-axis, and-1.85% for the unit cell volume. The a- and b-parameters as well as the orthorhombicity have a linear dependence on oxygen only in the underdoped range x = 6.5-6.75. The large number of measured compositions allows to measure the deviation from linearity. At x = 6. 75 the

40

E. KALDIS 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0

:"'

3,86: i i ~'

•

3,85T• >, str, 384- "~ E "~ o tra-

.

,-,

3,83-

=

Ed.

'N

;onder et al

. . . . . 13,86

~////~

~rageretal.

106:0.6:1 .6:2 6:3 6:4 6,5 6,6 6,7 6,8 6,9 7,0 0

~

l,t~///,/~

-

4384

e} 0 "F,

.

43,88

0 E

~

"%.

~

0 1:

381: (=~

'

......... 6,0'6:1'6:2'6',3'6',4'6:5'6',6"6:7'6:8'6:9'7,0'

~

'1381

0

/t-.~

~'/

Oxygen content x

'~

3884 i', o

T ~.

i,.~"" ' m1t3,8 0

~

E

,£

~,"

"

O

E

['hi

~

~ :

9 8

./I

r 388o

6 5

~, " '

4.

~..:.

4

.I. ,'= : ' :

3 2 1

3. 2 1 ........... o

6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 Oxygen content x

Fig. 25. Lattice parameters of equilibrium CAR samples as a function of oxygen nonstoichiometry, after data of Krfiger et al. (1997) and Conder et al. (1999). (a) a-parameter; (b) b-parameter; (e) orthorhombic strain; (d) c-parameter; (e) unit cell volume. Note the steep change at the T ~ O transition, and the deviationsfrom linearity.

6 5

-,85t. ~,

11,80

,,

1230 I ~j~3,865 Conde*retaL1999 .K:~igeret.al.,1997, .~3,860

..." •

tetragonal ~ / , orthorhombic O -1 . , , , . , . , . , . , . , • , . , -1 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 Oxygen content x

~"i

....

~

" :'

--~

E

/" •

......

70 11,90 J4 I~ L~. 11'85

o

~

o mm

~ ill0,=o ~ .~1 ~, ~

~" 11,70q~ g ! ] E ~ : tr'l~ ~t

"~

= ~"- ~ 1~'

]ToEi l= (~ 1175] '

~'3,865t El! ~ ~ ], ~; . . ~ 3,860 t ",2",':"i'i , ~ ,

.~., ~:

60 61 52 83 64 65 66 67 68 69 11,90 ] , , . , : . . . . . : . . . : . . . I . . . . : . . . . : . . . . : , , , , : , , , , I , T , ~ J : P- 123-Ox J : x ] : < Condereta 19891 I,=, KrOgeretal. 1997[

6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0

3,8.1

123-Ox

]3,85

.~///~/~ 997

.

9 8

""

11,80

I,,

I1~

I ~o

t o~ ] 1 1 7 5'

F"..Lo34.,70 / r"[~l

/

1185t , , ........ ,,..., . . . . /. .- . . ,....,....,...., . . . . . . . . ,11,65 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 Oxygen content x

6,0 6,1 6,2 6,3 6,4 6,5 6.6 6,7 8.8 6,9 7,0 177 . , . . . . . I . , . , . , . , . =177 insulator underdoped I [1 :"--.#i . superconductor I I ] 178- " : . . . . " . . . . . . [=1_4 176 4) =

•.,: -:3

: "~i

174

~: ~,: xz.

~ ~~-

E~: = &:

--"

173

~, '

(e)

o

175

T

175

O

~=

.~ -= o g_ ~=

0 ~

"-;~

I I 4 174

% L - ] .I

I 14173

-, , ¢. , . , . , . , , I i t 6,0 8,1 6,2 6,3 8,4 8,5 6,8 8,7 8,8 8,9 7,0 O x y g e n content x ,

.

,

.

x-dependence of both a- and b-parameters becomes nonlinear. As we will see, many other changes in properties appear at this composition, which is an important point of the phase diagram (sect. 5.4.2). Extrapolation of the tangent in the range x = 6.5-6.75 up to x = 7 shows the relative deviation from linearity to be +0.15% (Aa = 0.006 A) for the a-axis and appreciably stronger, - 0 . 2 5 % t a b = 0.010,~), for the b-axis. The c-axis of the CAR samples changes almost linearly with the O-content up to x ~< 6.95, opposite to the BAO and DO samples (sect. 3.2.2.4). We note that in contrast to the a- and b-axes it also does not show a change of slope at x = 6.75.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30x 11,906i0.6 1 6 2 . 6

t

3.6i4.6 5 6 6.6 7 6 8 . 6

i

11,85t

i

"~

Conder et al. 1999 Kr0geret al. 1997

i4

11,75 j t

o |#am =/ ° ~",m

11,70~

--/

t

11,65

~

I o_ I ~

11,8°t

11,8o 11,75

o ~i...

°/

(a)11,6516,0 , 6,1,

9.7,011,90

128-0

I, =,

~ = i m: ] o " " .....

41

11,70

°'

,. ,~'P. , . , , , . , , , . 11,65 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0

Oxygen content x

3,87............t: 3,86 i .=-=0=%•""" E

3'85 ~

"~

3'842

{11

3,83

1~

3,82 ~-

(b)

'"'""""""'i'"""" 123-O×

.... 3,87

!

3,86

Jorgensen et al. 1990a

3.85

°./

3,84

/%,,,

Conderetal.

8,83 []

' ". •

3,82

1999'. , ~=! KrLlger et ali1997 ",

3,81 .... ,....,....., ...................... .... i........... ". 3,81 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0

oxygen content x 6,6 -m:

6,7 .

"'..

.

.

,

8,8 . . . .

,

6,9 . . . .

7,0

....

underdoped iJ &'"m';.-,....

'E•174-

• m.,•

~

174

E

o

m

""11 O

e,.

(c)

173

123-0, Conder et al. 1999 KrlJger et al. 1997

6,6

6,7

173

6,8

6,9

Oxygen content x

. .1 7,0

Fig. 26. (a) Comparison of the c-parameter doping dependence of CAR samples (XRD 300K, solid squares; Conder et al. 1999, KNger et al. 1997) and quenched samples (NPD 300 K, open diamonds; Jorgensen et al. 1990a) both with Ba2CO3 precursor. The differences near the T-O transition are underlined with arrows. (b) Comparison of the a-parameter doping dependence of CAR samples (XRD 300 K, solid squares; Conder et al. 1999, Kriiger et al. 1997) and quenched samples (NPD 300K, open squares; Jorgensen et al. 1990a) both with Ba2CO3 precursor. The differences near the T-O transition are small, however, due to the large number of samples and the exact chemical analysis, the trends in the change of slopes are better visible for the CAR samples. (c) Decrease of the unit cell volume of the CAR samples in the optimally doped and overdoped ranges, reflecting the decrease of b-axis due to increasing interactions (Conder et al. 1999, Krfiger et al. 1997).

Other differences to fig. 12 appear in the o v e r d o p e d region. The unit cell v o l u m e shows an additional decrease in the optimally d o p e d and o v e r d o p e d regions (fig. 26c) reflecting changes o f the b- and, to a lesser extent, the c-parameter. The decrease o f the b-parameter clearly seen in the o v e r d o p e d r e g i o n (Kriiger et al. 1997), m i m i c k s the Tc d e p e n d e n c e on o x y g e n doping (fig. 25b). This shows that the structural changes due to ouerdoping w h i c h we discuss in sect. 6 and subsequent sections, are not only limited to the c-axis but are correlated with changes in the planes. M o r e p r o n o u n c e d anomalies appear in samples synthesized without carbonate, as we discuss in sect. 3.2.2.4.

42

E. KALDIS

1 10-5

I

0 -1 10-5

/"

"

I

I

I

I.

emu/g I

-2 10 .5

bB

x=6.396

-3 10.5 d5

FC 10 G

-4 10-5 -5 10.2 -6 10-5 -7 10-5

(a)

t

|

I

i

0

,

~

,

I

20

,

I

I

40

I

I

60

80

,

100

T (K) 0.00012

'

'

'

l

'

'

"

l

'

'

'

l

'

'

'

l

'

'

'

|

'

'

'

0

•o ZFC-100G I FC-100G 0.0001

:~ 8 1 0 -s

E

0

eo o

6 1 0 -5 o o •

o •

o •

o

4 10 -5

o

o

2 10 s

(b)

20

40

60

80

100

120

T (K) Fig. 27. (a) Diamagnetic transition clearly seen for 123-Ox with x=6.396. Field cooling, H = 1 0 G . After A. Shengelaya (Conder et al. 1999). (b) Small amounts of the sample become superconducting already at x = 6.342. H = 100 G, ZFC and FC. At H = 10 G and FC no distinct indication of superconductivity appears. After A. Shengelaya (Conder et al. 1999).

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTSIN YBa2Cu30~

43

The vertical lines (solid or dotted) in figs. 25 and 26 correspond to some of the important compositions (6.15, 6.4, 6.75, 6.90, 6.95) of the phase diagram correlated to changes of several physical properties, which we discuss in detail in other sections. Here, we discuss briefly some possible reasons leading to the nonlinearity of the a- and b-lattice parameters with oxygen content: (a) Partial occupation of the anti-chain sites 05. As already mentioned, slowly cooled samples investigated by high-resolution neutron diffraction (Radaelli et al. 1992, B6ttger et al. 1996), but also in quenched samples (Jorgensen et al. 1990a), an occupancy (5%) of the 05 site has been found. The relative increase of the a- and decrease of the b-parameter starting from x ~ 6.75 could be in principle explained by increasing 05 and decreasing 0 4 occupancies in the oxygen-rich regime (Krfiger et al. 1997). EXAFS investigations of these samples did not confirm up to now this assumption (R6hler et al. 1997a,b). Also single-crystal work (von Zimmermann et al. 1999) did not find 05 occupation. (b) Nonstoichiometry of Cul due to the incorporation of carbonate. However, the Rietveld refinement showed the occupancy of this site to be independent of the oxygen content, having values between 0.93 and 0.88 (Krfiger et al. 1997). The most probable reason for the deficiency of Cul is the carbonate traces of the order of 1000ppm (sect. 3.1.4) found in these samples. This assumes that one carbonate impurity would hinder the substitution of several Cul atoms. Cul site defects have been observed and discussed in detail in single-crystal refinements of the double chain compound 247-Ox (Y2Ba4Cu7014~) (Schwer et al. 1993a,b and references therein; Kriiger et al. 1994). In single crystals grown in alumina crucibles the main impurity is, however, Al-ions incorporated in the chains. Due to the higher temperatures involved, the carbonate incorporation is probably less important in single crystals. The CAR powder samples have been processed at lower temperatures (935°C) and, therefore, the main impurity is carbonate which, however, has been reduced to only ~1500 weight ppm due to an effective expelling procedure (sect. 3.1.2). The possibility for reaction with alumina is small at these temperatures. (c) The existence of a series of superstructures indicating long range interactions, which is discussed in sects. 5.1.1 and 5.5.2. (d) The change of the superconducting coupling. We conclude that the changes of the superconducting coupling in the optimally doped and overdoped ranges (leading to the decrease of the b-axis mimicking the Tc vs. O curve of fig. 2) are mainly responsible for the nonlinearity of the a- and b-lattice parameters. Several experimental results support this idea: As we will see most measurements influenced by the structure do show changes of slope at x ~ 6.75. Thus, e.g., the pressure dependence of Tc shows a giant anomaly in this range (sect. 5.2) and the onset of the diamagnetic transition shows an abrupt step in this range. This is the range generally considered as the Ortho II -+ Ortho I transition, but doubt must be expressed in view of the fact that recent superstructure investigations (von Zimmermann et al. 1999) find that at RT this transition takes place at higher oxygen content (sect. 5.1). Only at higher temperatures (T > 350 K) is the Ortho I phase present for x > 6.4.

E,

44 6,5 . , ....

11,78

E

ll,76

..

11,74 -

~,~

6,6 6,7 6,8 i .... i . : . . , ,i . . . .

•

""t 0

×RD, 3,99 K

11,66

,

i

i

.

11,78

. . . .

i

11,72

Ill

i'% J

Rusiecki and Kaldis 1991

11,68 -

,

11,74

" 11,70-

7,0 .;..

11,76

•

0

i

i

11,72 R

6,9

KALDIS

. . . .

6,5

6,6

6,5

6,6 ' ....

i

. . . .

6,7

~

11,70

i•

. . . .

6,8

i

z~-

#¢

,i,

,

6,9

,

11,68 i

'

11,66

7,0

Oxygen content x 175,2- ' ' '

....

175,0~'

6,7 6,8 6,9 7,0 ''i .... ~'''~'i •''''

•

!

!

i

==• ~i

!

i

174,8-

~ 174,6•

E 1744-

"~ 174,2i i

174,0- BAO samples

~ 173,8-

xRo,3o9,

173,6- Rusiecki and Kaldis

~ 173,4-

1991

..... ! i \ ~ii

i

i

i

i

i

i

=:

•

"% ==! i

~"

.t"

173,2 6,5

6,6

6,7

6,8

6,9

Oxygen content x

7,0

175,2 175,0 174,8 174,6 174,4 174,2 174,0 173,8 173,6 173,4 173,2

Fig. 28. (a) c-parameter of BAO-123-O~ samples (XRD, 300K) showing a minimum at x ~ 6 . 9 3 near the onset of the overdoped range. This was the reason to start the investigation for a phase transfonlaation (sect. 6). In contrast, the CAR samples show a linear behavior (cf. fig. 25d). (b) Unit cell volume of BAO-123-Ox samples (XRD, 300K) showing the minimum at x ~ 6 . 9 3 near the onset of the overdoped range• The anomaly in the x=6.73 6.80 range indicates the coexistence of two phases, in the range of the To,,,,l~ottransitionshown in fig. 2 (hatched). After data of Rusiecki and Katdis (1991).

3.2.2.4. Lattice parameters of BAO and DO samples. There are interesting differences among the lattice parameters of samples synthesized with (CAR) and without (BAO, DO) Ba carbonate as precursor (sect. 3.1.2). The absence of carbonate as a precursor of the solid-state reaction (DO) and the expulsion of carbonate (BAO) seem to render a particular sensitiuity of these samples to structural distortions (DO) and metastable phases (BAO). The effect is striking in the c-parameter which is almost linear for the CAR samples (fig. 25d), but shows changes of slope at x ~ 6.75 and x ~ 6.80, and a minimum at x ~ 6.93, for the BAO samples (Rusiecki et al. 1990, Rusiecki and Kaldis 1991) (fig. 28a). Corresponding anomalies appear for the DO samples as we will discuss later (Conder et al. 1994a, Conder and Kaldis 1995, Kaldis 1997). The minimum of the c-parameter at x ~ 6.93 corresponds to a displacive transformation at the onset of the overdoped region, which we will discuss in detail in sect. 6. As we have seen above for the CAR samples (sect. 3.2.2.3) in the range o f x ~ 6.75, the deviation from linearity of the in-plane lattice parameters - seen also in the orthorhombic strain and the unit cell volume - sets in. The corresponding change of slope for the c-axis is smoothed out in the presence of carbonate, but appears clearly in the BaO samples. Figure 28a shows also that for these samples the c-parameter seems independent of doping in the x = 6.73-6.80 range. A lattice parameter

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN YBa2Cu30x

45

independent of composition indicates a two-phase region, separating two different phases. It is consistent with the anomaly in the To,onset of fig. 2 which shows a clear step in this range. In this step, samples with To,onset~ 93 K and T~,onset"~ 74 K coexist, an indication of a two-phase region separating the two To phases (cf. fig. 45, below). Under the influence of the a- and c-parameters the unit cell volume (fig. 28b) shows this two-phase region more pronounced. However, a larger scattering of data appears in the curve of the unit cell volume due to the contributions of the in-plane lattice parameters, as we will discuss below. Compared to the c-axis the in-plane lattice parameters of the BAO and DO samples show a larger scattering. They show deviations from linearity but the situation is more complex. In the first case (BAO) the number of investigated samples (45) in the 6.5-7.0 range is as large as that of the CAR samples, giving good statistics. Thus, in spite of a larger scattering of the data the changes of slope can be clearly seen. Figure 29a shows the raw data of the a- vs. x plot fitted with a simple polynomial. As this, however, is a physical problem and not a statistical one, treatment of the data in three nonstoichiometry segments, as suggested by figs. 28a,b, seems reasonable. This is shown in fig. 29b. The linear fitting in three segments covers 75% of the samples (solid line). The remaining 25% have higher a-values and lie on a parallel curve (dotted line). They are probably from samples pinned in a nearly metastable state, not being able yet to reach equilibrium. Upon closer inspection of the data it can be seen that the horizontal part at the transitional range near x = 6.75 (cf. sect. 5.4.2) appears also in other composition ranges, as shown in fig. 29c. Assuming a reproducibility error of 0.0025 A in the a-parameter (as compared to the fourth decimal point in the CAR samples, Krfiger et al. 1997), due to the metastability effects" and the reproducibility of the BAO method, it is possible to include almost all data in a tentative scheme of a series of four phases separated by two-phase regions (hatched). To avoid misunderstandings we stress that the coexistence of phases is not supported by two-phase samples. All samples are macroscopically (XRD) phase pure. It is the scattering of the a-values in certain x-ranges which tentatively implies that some of them are pinned in another metastable phase. The figure is drawn in the frame of a classical thermodynamic picture, but other models could be also applicable. Inspection of figs. 28a and 29 shows - as mentioned ah'eady - that the scattering of data is present mainly in the in-plane lattice parameters, particularly in a-, which are sensitive to oxygen ordering (superstructures). Due to the lack of exhaustive characterization of the BAO samples, fig. 29c might be considered as tentative. It cannot be ignored, however, that it is qualitatively consistent with a sequence of superstructures as indicated by Raman investigations (sect. 5.5) or by recent hard X-ray investigations (sect. 5.1). This stresses the importance of systematic comparison and evaluation of the reaction mechanism of samples prepared by various methods. The yet unknown mechanisms of the various solidvapor-liquid reactions (in mesoscopic and atomistic scale) involved in the preparation of these complex perovskites seem to be specifically sensitive to metastable and stationary states. Some, possibly under the influence of grain-interface wetting by a flux (e.g., carbonate traces at high temperatures) homogenize the reacting batch macroscopically (CAR method, sect. 3.1). Other synthesis reactions using BaO precursor (BAO method,

46

E. KALDIS

3,845 - -

6,5 ,

6,6 ,

,

• 3,84O-

~.~-

~--~'~3 835,

•

•

6,7 ,

.~ .

.

6,8 ,

6,9 ,

.

7,0 ,

.

Fit of DatRusiecki_a

Y =5,26158=0,38295X+0,02528X2 .

•

6,5 3,845 3,840

E3,890

3,820-

6,7

XRD,300K

6,8

1991

3,885.

6,9

i....i. ' . = "

=.

a

l

£

t

...... ~ / " .,

7,0

"!' . . . . . . . .

.....! . . . -

Rusiecki and K 3,835 ~

3,830-

6,6

3,895 . BAO's'am'ples . . . . .

3,895

~ [

s

3,89O

~

.

i="

3,885

3,830

123-Ox BAO samples

-"

3,820

1~ 3 , 8 7 5

3,875

.la

3,815 (a)

,

.

8,5

3,845

6,5 ,

,

•

6,6

.

6,6 ,

,

.

,

.

6,7 ,

,

6,8 ,

,

.

,

6,9

.

6,9 ,

3,815 (d)

7,0 7,0 ,

.

123-0x, BAO Rusiecki and Kaldis 1991

""..

3,830.

3,825

~ " 3,820.

3,820

3,815

(b) 3,845"

, 8,5 6,5

3,840-

~

~,~ 3,835.

~

,

, 6,6

•

, 6,7

.

, 6,8

0 content x

6,6

6,7

6,8

, 6,9

,

, 7,0

6,9

7,0

123-Ox, BAO samples Rusiecki and Kaldis 1991

~

-. ~

• ,. ............

3,820-

__

3,815 6,7

6,8

~.

6,7

6,8

6,9

(e)

3,870

7,0 9

''"87 t 6

64 O

,., .... 6,5

, .... 6,6

,.i,.,.,.,., 6,7 6,8

....

, . 7,0

6,9

Oxygen content x

6,60 6,65 6,70 6,75 6,89 6,85 6,90 6,95 7,00 ,, .... , .... , .... ,.;...,,. , .... , .,.~...

11671

3,830

E

,

i

:,

,: :: ~

11,65 11,64

. (f)

!

i

i

11,83

6,60 6,65 6,70 6,75 6,80 6,85 6,90 6,96 7,00

oxygen content x

3,884

6,60 6,66 6,70 8,75 6,80 6,85 6,90 8,95 700 , 3,884 ,

i

.

,

. 123-O 3,883- BAO samples 3,882 - A.Hewat et al. 1991 •

,

r

=

i

i _ /:: •

,

_

j

,

~

,

i

,

: i

'

3,883 ~

3,882

i

. ~ 3,881-

X

11,86

i

A.14De'wTa~3e~ OK 1991

0 content x

Fig. 29. XRD data at T = 3 0 0 K : (a) a-parameter of BAO-123-O= samples vs. x, The solid curve is a polynomial fit. (b) The scattering of the data of fig. 29a indicates the existence of the three segments of x appearing in fig. 28a, with a stepwise decrease of the a-parameter with increasing oxygen doping. (c) Possible two-phase regions separating a sequence of phases (superstructures) (cf. figs. 65, 110, 111). The arrows indicate the stability region of each phase. For a discussion see text. (d) b-parameter vs. x. (e) Orthorhombicity vs. x. After data of Rusiecki and Kaldis (1991). NPD data at T = 5 K: (f) c-parameter. The minimum found by XRD is confirmed, The lower values are due to the thermal contraction at low temperatures. (g) b-parameter vs. x. The lines are guides to the eye. After data of Hewat et al. (1991).

11,67

T ~",,.~ } i \

1123-Ox 11 654 BAO samples

3,820

7,0

i

11,68

3,815 6,9

6,6

,,,, 7,0

6.

3,845

3,825

I 6,6

, ...... 6,8 6,9

3,840

~!'i ~,,~. ,~

~3,825. 6,5

3,815

~

3,835

~- 3,830

(C)

.

, .... 6,7

Oxygen content x

6,5

3,830

3,825.

, .... 6,6

9" BAO samples i . ~ XRD 3OO K, . .~,~-• : " " ~Rusiecki and ~/~" . .~... X Kaldis 1991 ~=...,~.~ " ''' .......... '' ' .... " ..... .~-7

3,840

i--

3,870 . . , . . . . 6,5

3,845

samples

• =.,~, 3,840

•

O6'7content6'8 x

.

3,881

3,880

3,880

3,879-

3,879

(g) 3,878

, . , . , , , i , . , . . , . 3,878 8,60 6,65 6,70 6,75 6,80 6,85 6,90 6,96 7,00

oxygen content x

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN YBa~Cu3Q

47

sect. 3.1) are easier pilmed (partly) in metastable states. Similar is true for the DO method using direct oxidation of the barium metal (sect. 3.i). Figures 29d,e show the b-lattice parameter and orthorhombicity behavior of the BAO samples. In spite of the scattering of the data in the oxygen-rich range the decrease of the b-parameter in the overdoped range can be seen in fig. 29d. The DO method (cf. sect. 3.1.2.1 and table 2) has its own fingerprint of structural anomalies, indicating a structural-pinning reaction mechanism. In spite of many grindings and annealings, reaction and phase transformation rates are probably slower, pinning the structure in metastable states. The number of samples investigated was limited (20), but, owing to the use of a better controlled closed system with higher reproducibility of the synthesis conditions, the scattering of the data was smaller than for the BAO samples and could give some additional information about the overdoped range. Figures 30a-e show the lattice parameters vs. x of the DO samples. The spikes of these curves seem at first glance to be the result of scattering. Closer inspection shows that they coincide with the five characteristic compositions occurring at x ~ 6.7, 6.8, 6.9, 6.95, and 6.97. As mentioned for the BAO samples, the anomalies in the 6.7-6.8 range do appear also for T~,o~,s~t(fig. 2, hatched area) and are, therefore, associated with the superconductivity. We note that the CAR, BAO and DO curves are linear and do not show any scattering up to 6.7 (up to 6.8 for the a-axis). Only slight differences in slope exist in this range. At x ~ 6.7 however, the b-axis drastically changes its slope. At the onset of the optimally doped range (x = 6.90), all parameters (particularly a-) show a maximum. The same is true for the displacive transition at 6.95. Even the small anomaly at 6.97 can be attributed to an anomaly of the dimpling (R6hler et al. 1998). Last but not least, it is very interesting to compare the behavior of the DO samples with that of the BAO samples with respect to the c-axis anomaly (fig. 30d). We then find that the baseline of the DO curve coincides with that of the BAO curve (dotted line) in the range x > 6.80. In view of the fact that method of synthesis, starting materials, apparatus, and operator were quite different we can conclude that this is not a mere coincidence, but these two methods do really reach states which are not attainable for the CAR samples (different activation energies), whose c-parameters remain higher and change linearly with x. In addition, the DO method shows characteristic spikes in the compositions of the above mentioned transitions, supporting the assumption of structural pinning. Summing up, the lattice constants of near-equilibrium samples synthesized by three different synthesis methods indicate the possible existence of several two-phase regions (miscibility gaps), and transitions in the whole nonstoichiometric range (x = 6.0-6.99). We list them here: (1) x=6.3-6.45. CAR samples: structural T-O transition. It appears in this range only for the in-plane lattice parameters. For the c-axis it is much more abrupt and appears at x-- 6.40 (-6.45), coinciding with the onset of superconductivity. (2) x=6.75-6.80. CAR samples: deviation from linearity of the a- and b-axis. BAO samples: change of slope of all lattice parameters; c and V independent of doping, indicating possible miscibility gap. The DO samples show also changes of slope. In

48

E. KALDIS 6,4

6,5

6,6

3,840 • ' . . . . . . . . .

6,7

. ........

6,8 ,

......

-~3,835

t

3,820. 3,815 . . . . . . . . . 6,4 6,5

1

~

-

-

t

d°Pedl : I =4 i~ I

r- \

21

11,66

'E

l

1

t

I

H

/

1,67

3,825

Conder and Kaldis 1995 Conder et al. 1999

(a)

7

-/L oo~o,do~ ;~'-l'ove,d,,.I-

11,694

t

123-O, DO samples

~ 3,825- XRD 300 K

,

t

,.

3,830

3,830-

1 ~.

I ~. / II ~1.701 v II

. . . . . . . t3,840

3,835-

i

1

7,0

6,9

. ......... 6,6 6.7

, ..... 6,8 6,9

::kid

3,820

(d)

1

1 6,75

, 6,80

6 6 ~ 6,85 6,90 6,95 7,00 Oxygen content x

6,70 6,75 Z \ /

'

6,80

6,85

6,£0

7,05

6,95 7,00 overdoped1744

Oxygen content x

lit° t3090

64 65 66 67 68 6,9 3,890 '1 . . . . " . . . . : . . . . : . . . . " . . . . . . DO samples !~ J 3 888, XRD300 K y~ ~ ~ J ' Cender and K a l d i s ~ T

3,886 ~96

Conder et al. /

y

::

~

-~ 3,880-

=

Iconuerelal. 1~,Bl1998 199

3,886

i:

ii

t 3,884

t ;~::<,o,e,~,.

- .....]'°°°

:: /

..... .... l

[

&70

(e) 3,878 . . . . , . . . . , . . . . .. 6,4 6,5 6,6 6,7

(b)

,•',.,,,, 6,8 6,9

6,5

6,6

6,7

6,8 i

o

"~

e-

8,5- XRD, 300 K Conder and Kaldis 8,01995 ./~ 7,5- C o n ~

6,9. . . . . 7 , 0 ~1/

~1"

2

7,0~

6,5~ 6,0 5,5

0

(c)

5,0~ 6.4

................ 6,5 6,6

6,7

.., .... , ...... 6,8 6,9 7,0

Oxygen content x

~

: "~ i : ~

2

~.Jl'/

~! / ,~/ m ~-='

/:

: i :

/ ~1~,8

t

'<.l.t t ' 1 .....

I l'-41(;.t . . . . . . . &75

::

/

7"1.

"m

'

.....

I. S.95F! I7,00....

6,80 6,85 o.~o Oxygen content x

7,0

Oxygen content x 6,4

:

l DO-saranlos

:

/

e 3,882

174,0

'l

7 9,0 4 8,5 8,0 7,5 7,0 6,5 6,0 7 5,5 '5,0

Fig. 30. DO, like BAO samples, show typical anomalies of their lattice parameters at x ~ 6.75, 6.80, 6.90, 6.95 and 6.97. These anomalies seem to be sensitively dependent on the existence of latent structural and electronic transitions. For a discussion see the text. XRD, T = 3 0 0 K : (a) a-parameter vs. x. (b) b-parameter vs. x. Note that only the b-parameter changes slope at x ~ 6.75 (cf. sect. 5.4.2). (c) Orthorhombicity vs. x. (d) c-parameter vs. x. Note the overlap of the DO data with those of the BAO (dotted line). In the 6.97 >x > 6.75 range, both are appreciably smaller than the CAR data for the c-parameter (black circles). (e) Unit cell volume vs. x. After data of Conder et al. (1994a, 1999) and Conder and Kaldis (1995).

this range many changes of physical properties take place: transition of the T~ plateaus 60 --+ 90 K takes place, cf. fig. 2 (see also sect. 5.4.1); change of slope in uniaxial and hydrostatic pressure experiments, giant pressure effect of To (sect. 5.2); steps in the

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN Y B ~ C u 3 0

x

49

in-phase phonon Raman shifts appearing at the compositions of nao superstructures (Rarnan, sect. 5.5). (3) x ~ 6.50--6.99. a-axis BAO samples, indications for the existence of a sequence of phases separated by miscibility gaps. This would be consistent with the appearance of superstructures. (4) x = 6.90. DO samples: clear anomaly (spike) of the curves vs. x, coinciding with the onset of the optimally doped phase. (5) x ~ 6.95. BAO samples: c-axis minimum, coinciding with the displaeive transformation at the onset of the overdoped phase. Triggered by these results and NPD investigations (discussed later), EXAFS measurements showed a first-order-like jump increasing the dimpling Cu2-O2,O3 (Kaldis et al. 1997b, R6hler et al. 1998), the Y-O2,O3 distance becoming smaller (sect. 6.5). As we will discuss in sect. 6.3, corresponding anomalies have been found in Raman measurements. (6) x = 6.95. DO samples: spikes of the curves vs. x, coinciding with anomalies of the dimpling (cf. sect. 6.5) (R6hler et al. 1998). The conclusion is that the anomalies of the lattice constants -particularly for samples

synthesized without carbonate - correspond to important changes of the lattice, some of them clearly related to superconductivity. Carefully measured lattice constants in highquality equilibrium samples combined with a high-resolution determination of oxygen is a very sensitive probe for phase boundaries and transitions. We recall that due to the high resolution of the volumetric oxygen determination (Conder et al. 1989) it is the first time that the dependence of the lattice parameters on the oxygen nonstoichiometry can be investigated so exactly and even small anomalies can be detected. Tables 5, 6 and 7 give the values of the XRD lattice parameters at T = 3 0 0 K (Guiifier camera, with Si standard) for the CAR, BAO and DO samples, respectively. To confirm the anomalies in the lattice parameters found by XRD, an NPD refinement was performed as a function of temperature at 12 oxygen contents. Tables 8a-d give neutron diffraction data of 12 selected BAO-sample compositions at T = 5 K and 300K (Hewat et al. 1991). Figure 29f shows the c-parameter vs. x as measured at 5 K. Both anomalies found with XRD were confirmed, as apparent upon comparison with fig. 28a. The same is true for the b-parameter shown in fig. 29g, where the decrease in the overdoped range is seen more clearly than in fig. 29d. The NPD bond lengths of the BAO samples are discussed in more detail in conjunction with the dimpling transition corresponding to the minimum of the c-axis (sect. 6.1). It is interesting that although the CAR samples do not show the c-axis anomaly at the large length scale of XRD, Raman and EXAFS investigations, which have a much smaller length scale (sect. 6), find the corresponding anomalies at RT and T = 25 K, respectively. We may conclude, therefore, that the corresponding structural changes appear in the CAR samples at mesoscopic scale. Therefore, the chemical and thermal history of the samples can mask several structural properties at the macroscopic scale, which, however, can be still detected with small length scale experimental methods. This is an important facet of the complexity of the HTc cuprates: the different properties between micro- and macroscopic scales.

50

E. KALDIS Table 5 Lattice parameters of CAR samples ~ O content

a

b

6.03

3.8611

3.86ll

6.066

3.8586

3.8586

6.12

3.8612

3.8612

6.116

3.8595

6.143

3.8604

6.172

3.8603

6.286 6.342

Orthorhombicity

c

Volume

1

11.8417

176.53717

1

11.8337

176.18952

1

11.8285

176.34951

3.8595

1

11.8299

176.22

3.8604

1

11.8266

176.25

3.8603

1

11.8237

176.2

3.8614

3.8614

1

11.8052

176.02037

3.853

3.868

1.94275

11.807

175.96449

6.396

3.8469

3.8724

3.30341

11.793

175.6772

6,442

3.8406

3.8739

4.31655

11.7466

174.76

6.482

3.8397

3.8782

4.98841

1.7432

174.87

6.53

3.8346

3.8787

5.7174

11.7437

174.63

6.53

3.8351

3.8791

5.70377

11.7407

174.63

6.563

3.8329

3.8804

6.158

11.7367

174.56

6.569

3.833

3.8808

6.19669

11.7361

174.58

6.613

3.8307

3.8828

6.7544

11.732

174.46

6.657

3.8276

3.8838

7.28791

11.7254

174.31

6.668

3.8277

3.8844

7.35208

11.7245

174.32

6.67

3.8276

3.8842

7.34

11.7268

174.35

6.671

3.828

3.8846

7.34

11.7229

174.32

6.7

3.8268

3.8853

7.5854

11.7213

174.27

6.709

3.8256

3.8851

7.7165

11.7176

174.16

6.755

3.8242

3.8868

8.1381

11.7136

174.11

6.762

3.8246

3.8868

8.06598

11.7148

174.14

6.778

3.8242

3.8869

8.13

ll.7101

174.06

6.781

3.8234

3,8872

8.27

11.711

174.05

6.806

3.8227

3,8876

8.41731

11.7085

174

6.82

3.8216

3.8878

8.58692

11.7064

173.93

6.826

3,8223

3,888

8.52107

11.7041

173.92

6.853

3,8211

3.888

8.68

11.7046

173.89

6.857

3,821

3.8881

8.7

11.701

173.83

6.886

3.8201

3.8881

8.82177

11.6968

173.73

6.888

3.8198

3.8877

8.8096

11.6958

173.69

6.909

3.8192

3.8874

8.84956

11.6926

173.59

6.912

3.8194

3.8875

8.83624

11.6923

173.61

6.94

3.8189

3.8873

8.88

11.6919

173.57

6.94

3.819

3.8868

8.8

11.6909

173.54

6.967

3.8181

3.8854

8.74

11.6821

173.31

6.968

3.8187

3.8856

8,684

11.684

173.36

con~nuedon nextpage

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS 1N YBa2Cu30~.

51

Table 5, continued O content

a

b

Orthorhombicity

c

6.981 6.981 6.981 6.983 6.983 6.984 6.984

3.8179 3.8179 3.8177 3.8177 3.8174 3.8182 3.8175

3.885 3.8848 3.8849 3.885l 3.885 3.8852 3.8851

8.71 8.68527 8.72433 8.75006 8.77649 8.7 8.77626

11.6825 11.6827 11.6828 11.6835 11.6829 11.685 ll.6841

Volume 173.28 173.28 173.27 173.29 173.26 173.34 173.29

Conditions of synthesis with carbonate precursor, sect. 3.1.2). XRD at T=300K. STOE Diffractometer, use of silicon standard. Data after Krfiger et al. (1997) and Conder et al. (1999). 3.3. The T--+ 0 transition The difference between Zr-gettered samples quenched in air from lower temperatures (Cava et al. 1987a,b, 1990) and quenched in liquid nitrogen from higher temperatures (Jorgensen et al. 1990a) impeded in the beginning the understanding o f the nature o f this transition. The T ---+O transition was implicated in the argument on abrupt change or continuous change of the lattice parameters, the former being considered as an indication for a first-order transition and the latter as an indication for a second-order transition. The two different methods o f synthesis are compared in table 4. A comparison o f the lattice constants is given in fig. 12. The same difference appears also by the contraction o f the apical bond which is correlated to this transition (fig. 14). As we have seen, the abrupt change o f the lattice parameters at x = 6.35 has been confirmed with equilibrium C A R samples (Conder et al. 1999, fig. 25). On the other hand, NPD at high temperatures by Jorgensen et al. (1987b) indicated that the transition is second order. A drawback of the work o f Jorgensen et al. (1990a), as in all quenching experiments, concerns the effectivity o f quenching which is never instantaneous and complete. Thus, a certain disorder will always remain in such samples. The low temperature o f the quenching in air used by Cava et al. and the relatively slow gettering with zirconium probably succeeded in bringing these samples not far from thermodynamic equilibrium. Using such samples Cava et al. (1987a, b, 1990) could support and extend the Tc plateau found by Tarascon et al. (1987a). The latter reported, based on first ac susceptibility measurements, that samples prepared in flowing gas were a mixture o f To ~ 91 and Tc ~ 55 K materials. The important contribution o f Cava et al. (1987a,b) at this point was to show clearly the existence o f the two "plateaus" in the Tc vs. x curve. However, some questions were raised by this work. Although their samples extended into the range x ~> 6.30, their lattice constants did not show the I - M transition, all the samples staying superconducting (Tc ~ 27 K at x = 6.3). In a later more systematic investigation, Cava et al. (1990) gettered their samples at the fixed temperature o f 440°C and varied the stoichiometry by changing the amount o f Zr foil (table 4). The superconductivity

52

E. KALDIS Table 6 Lattice parameters of the BAO samples (Ba-oxide precursor, sect. 3.1.2) ~ O content

a

b

Orthorhombicity

c

Volume

6.517

3.841

3.879

4.92

11.74

174.91

6.52

3.837

3.878

5.3

11.75

174.84

6.531

3.839

3.875

4.7

11.751

174.81

6.546

3.842

3.876

4.4

11.76

175.13

6.577

3.841

3.873

4.15

11.764

175

6.58

3.836

3.881

5.8

11.739

174.76 174.58

6.592

3.834

3.88

6

11.736

6.613

3.832

3.884

6.74

11.725

74.51

6.634

3.832

3.882

6.48

11.723

174.39 174.34

6.638

3.829

3.883

7

11.726

6.641

3.832

3.883

6.61

11.726

174.48

6.65

3.833

3.883

6.5

11.728

174.55 174.43

6.654

3.83

3.884

7

11.726

6.673

3.83

3.885

7.12

11.72

174.39

6.675

3.831

3.885

7

11.725

174.51

6.702

3.829

3.886

7.4

11.718

174.36

6.707

3.834

3.886

6.74

11.714

174.53

6.714

3.834

3.886

6.74

11.713

174.51

6.734

3.825

3.888

8.2

11.709

174.13

6.748

3.829

3.887

7.52

11.715

174.36

6.768

3.826

3.89

8.3

11.706

174.22

6.795

3.826

3.889

8.15

11.713

174.28

6.808

3.826

3.892

8.55

11.695

174.15

6.81

3.83

3.886

7.26

11.7

174.14

6.813

3.825

3.89

8.42

11.703

174.13

6.822

3.827

3.892

8.4

11.707

174.37

6.828

3.822

3.89

8.8

11.706

174.04

6.833

3.831

3.888

7.38

11.703

174.32

6.851

3.823

3.889

8.56

11.692

173.83

6.862

3.821

3.889

8.82

11.682

173.59

6.875

3.828

3.888

7.78

11.685

173.91

6.875

3.826

3.888

8.04

11.684

173.81

6.876

3.822

3.89

8.81

11.691

173.82

6.888

3.824

3.893

8.94

11.689

174.01

6.892

3.822

3.889

8.7

11.677

173.56

6.902

3.822

3.891

8.95

11.682

173.73

6.906

3.822

3.889

8.69

11.68

173.61

6.92

3.825

3.886

7.9

11.668

173.42

6.932

3.822

3.887

8.43

11.672

173.4 continued on next p a g e

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBazCu30 x

53

Table 6, continued O content

a

b

6.938

3.821

3.888

6.939

3.821

3.888

6.954

3.82

3.888

Orthorhombicity

c

Volume

8.7

11.671

173.38

8.7

11.673

173.42

8.8

11.674

173.38

6.955

3.819

3.888

8.95

11.677

173.38

6.96

3.82

3.888

8.8

11.676

173.41

6.967

3.82

3.888

8.8

11.675

173.4

6.973

3.82

3.888

8.82

11.679

173.46

6.982

3.819

3.888

8.95

11.686

173.52

XRD at T = 3 0 0 K (Guinier camera, with Si standard). After Rusiecki and Kaldis (1991).

Table 7 Lattice parameters of DO samples (Ba-metal precursog sect. 3.1.2)" O content

a

b

Orthorhombicity

c

Volume

6.48

3.8388

3.8797

5.29896

11.741

174.86332

6.556

3.8362

3.8827

6.02417

11.732

174.74595

6.648

3.8325

3.886

6.9314

11.7207

174.5575

6.699

3.8299

3.8879

7.51509

11.7162

174.45736

6.738

3.8283

3.8886

7.81402

11.7101

174.32507

6.789

3.8265

3.8884

8.02344

11.6992

174.07196

6.795

3.825

3.8874

8.09087

11.7004

173.97682

6.813

3.8248

3.889

8.32275

11.7092

174.17022

6.821

3.8235

3.889

8.49271

11.6939

173.88352

6.831

3.8227

3.8889

8.58447

11.6954

173.86496

6.881

3.8243

3.8884

8.31097

11.6737

173.59268

6.903

3.824

3.8885

8.36305

11.6813

173.69654

6.913

3.8281

3.8891

7.90442

11.6718

173.76817

6.922

3.8255

3.8863

7.88402

11.666

173.4389

6.947

3.8201

3.8879

8.79606

11.675

173.39905

6.947

3.8211

3.8883

8.71663

11.6717

173.41325

6.948

3.8222

3.8877

8.49557

11.6801

173.56123

6.964

3.818

3.8843

8.60782

11.681

173.23224

6.968

3.8205

3.8868

8.60223

11.676

173.38299

6.97

3.8183

3.8857

8.7487

11.6807

173.30384

6.974

3.8195

3.8864

8.68166

11,6847

173.44891

6.986

3.8176

3.8866

8.95615

11.6834

173.35226

XRD at T = 3 0 0 K (Guinier camera, with Si standard). After Conder and Kaldis (1995)

54

E. KALDIS

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

~

.

~

~

~

~

~

~< v ~

cq

%

?

c~

c~

z

o

O X Y G E N NONSTOICHIOMETRY A N D LATTICE EFFECTS IN YBa2Cu30 ~

55

,6

~6

,.6

©

,.6

g eel II

r-i

,.6

r...)

e,3

~4~

~

tr3

tm

oo

o~

oo

~-

,,6

,.6 v

,.6

-g

<

56

E. KALDIS

o'h

,5

v

~

d d d d d

~

~ m

~M NMN~~ N M

~

d d d d d

~

.

.

.

.

o< ,,d

M

~~ N. N

.

~.

.

N

oO

,,6

'17, ,.,d o~ r.~ >..,

,,.d

Z

',O

~

.

.

.

.

,,.4,,

0",

,6

X X

X

X

X

¢q X

F~ r)

<

OXYGEN

NONSTOICHIOMETRY

AND

LATTICE

EFFECTS

'q.

o ,.6

~ m

M

M

M

M

~

~

.

.

.

.

~

M M M N ~

~

.

.

.

.

~

M M M N ~

~

.

.

.

.

MM

~ M M M M

~

.

.

.

.

MM

M M N M ~

M

~

~

M M M N N

~

.

.

.

.

~

M M N M M

m

.

.

.

.

~

~

~

t",l

~

o~ ,,6

c.o ,,6

t¢3 ,6

oo

II ,,6

C~

~

,.6

tt-3

z

~6

~

C',I eel ~

"~f

~

~

t"l

eel

~,~

N

"~

<

1N YBaeCu30 ~

57

58

E. KALDIS

disappeared at x ~ 6.35. It is a pity that the 27 K samples were not further reproduced and investigated parallel to the other samples, because a comparison would have extended our knowledge of how the solid-state reaction influences the superconducting properties. The stepwise change of the apical bond (fig. 14) and of the BVS (fig. 21) at the transition supported the idea (Miceli et al. 1988, Cava et al. 1988a,b, 1990) that the I-M transition is coupled with the T - O transition enabling the onset of the charge transfer.

3.3.1. The structural T---~ 0 transformation at the insulator-superconductor transition The order of the T---+O transition has been the object of many discussions. Thus, the short-range interactions ASYNNNI model (sect. 5.2) (de Fontaine et al. 1990a,b) predicted a first-order transition and a miscibility gap at intermediate temperatures. The temperature and composition dependence of the T - O transition it predicts is in very good agreement with experiment. On the other hand, Monte-Carlo simulations (Aukrust et al. 1990) and transfer matrix scaling (Bartelt et al. 1989) predict second-order phase transitions in the whole phase diagram. Only for an Ortho II-tetragonal phase transition at zero K they predict a first-order transition. We recall that in the first-order structural transformations the lattice deformation triggering the change of structure takes place via a large-scale motion preserving the topological integrity (lattice correspondence). A homogeneous strain drives the lattice transformation, the atoms moving in a cooperative fashion with speeds which can reach the speed of sound (dispIacive transformation). The difference to the diffusive second-order reconstructive transformations, where bonds to the neighboring atoms are broken and re-formed, is well known: the individual atoms diffuse to a locally favorable environment. The appearance of a "tweed" microstructure near the transition temperature may suggest a martensitic transformation, which in most cases is a first-order structural transition. Such transitions are very common in low-temperature (LT) superconductor materials (Krumhansl 1992) like V3Si and Nb3Sn or in ceramics and alloys undergoing chemical phase separation. An example of a displacive martensitic transformation is the square ~ rectangular lattice. A two-dimensional analog is the T---+O transition in 123-Ox. In the basal plane of the perovskite structure the number of available oxygen sites (04 and 05) is twice the number of oxygen atoms (04). Both sites are statistically occupied in the tetragonal phase. When these O atoms break the twofold symmetry between the sites and occupy only the b-axis sites (04) the T ---+O martensitic transition takes place (Kartha et al. 1995). In spite of their first-order nature, such transformations show pretransitional effects up to hundreds of degrees above the transition temperature. Figure 31 shows schematically this precursor regime. The tweed pattern of 123-Ox has been studied very well with highresolution electron microscopy (HREM). According to Kartha et al. (1995) the driving force for the appearance of this microstructure is the disorder of the material as it nears the transformation.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30 ~

59

T W E E D (I0-I00 A)

'-~ E

~- ~°8°-

~.2

T(transf)

o TWINNING (300-I000A) E

Fig. 31. Schematic diagram of the temperature fields of the tweed precursor of a martensitic transformation and of the twinned lowertemperature phase. Tweed has mesoscopic dimensions.

Although not directly connected with the 123-Ox case, it is worth mentioning here the case of Fel xPd,, showing the extreme sensitivity of the martensitic transformations to compositional changes. The martensitic transformation for x - 2 9 % appears at room temperature. Increasing the Pd content by 3% suppresses the transition temperature. A change of 1% in composition reduces the temperature by 100K[ This illustrates impressively how easy it is to miss a transition or misinterpret its order by assuming continuous change of e.g., the lattice parameters, if the compositions very near to it are not investigated (sect. 6). Following the ideas of Krumhansl (1992) and Kartha et al. (1995), the role of disorder in creating pretransitional phenomena can be illustrated easily with the above-mentioned Fe-Pd alloy. During the cooling of the alloy a local variation of the Pd content results, due to simple disorder quenched-in from higher temperatures. Small Pd-poorer regions will tend to transform to the martensitic phase already at higher temperatures. The statistical distribution of such mesoscopic domains leads to pretransitional deformations accompanying the first-order transition. Thus, the tweed theory of Kartha et al. (1995) treats the overall system as a collection of local regions interacting via the extended strain fields produced from these deformations. This cooperative behavior and the fact that tweed is not a transient state, because the disorder is always existing, are the differences to other models used earlier for the tweed of 123-Ox (Semenovskaya and Khachaturyan 1993). Novel theoretical models (e.g. Bishop 2000), give a general frame for the importance of the elastic deformation energy which, developing long-range forces, might be more important than the Coulomb energy for HTo superconducting oxides. Intrinsic nanoscale patterning of spin, charge and lattice degrees of freedom is considered as an important effect, which may lead to the coexistence of inhomogeneous superconductivity and magnetism. The difficult task to find some indications that at the macroscopic scale the transition is first-order has been undertaken by Radaelli et al. (1992) using ErBa2Cu3Ox (Er-123-Ox) slowly cooled samples, equilibrated at low temperatures (table 2). The reduction was made by equilibrating a mixture of x = 6.9 and x = 6.1 samples (cf. sect. 3.1.2.2). As do most

60

E. KALDIS 3.96

'

3.94

I °O

'

I

'

I

'

I

'

I

,

I

,

I

,

i

,

I

1

,

,

i

,,

,

i,

t

,

,

,

J ,

i

,

20 x103

i

o

o 3.92

E

0.8

16

"'

0.6

12

"~

8

u .~ E o

¢,,-

3.90

2 o 0.4 >U c 0~ 0,2

O 0 0 O O 0 0

3.88

, ~ e o o

•

b

•

3.86

Z

3.84 3.82

"::?,,,

.,.,

0.0

0.2

0.4

0.6 ×

0.8

4

u U

o

.0

Fig. 32. Lattice parameters of Er-123-O0+,. as a function of oxygen nonstoichiometry. Single-phase Rietveld refinements, solid signs; two-phase refinements, open signs. The latter method shows a discontinuity of the c-axis. After Radaelli et al. (1992).

o ¢.t..t

0

0.0

I

0.2

,

I

,

I

0.4

,

I

i

0,6

I

~ I

,

0.8

I

0

1.0

o

X

Fig. 33. Orthorhombic strain and occupancy order parameter as a function of oxygen nonstoichiometry. Extrapolation of the orthorhombic strain to zero gives the compositionof the transition. The discontinuity of the ordcr parameter indicates the first-order character of the transition. After Radaelli et al. (1992).

investigations in this field, this one suffers from an only approximate determination o f oxygen by neutron diffraction ( ~ 4% accuracy). However, in this particular case this is not jeopardizing the results concerning the order o f the T --+ O transition. In fact, the work o f Radaelli et al. (1992) is a highlight o f diffraction work in the HTc field. Its importance is augmented by the fact that it illustrates once more how important high-quality samples are for disentangling the properties of the HTc superconductors. Their structure refinement with NPD showed clearly a systematic broadening o f the diffraction peaks near the T ---+O transition, supporting earlier findings (Jorgensen et al. 1990a,b). This broadening shows two important characteristics: - anisotropy, i.e. hkl dependence, as has been shown before (David et al. 1989), and - asymmetry o f the line shapes, which is independent of strain or particle size. Agreement between calculated and experimental diffraction patterns was only achieved when both an orthorhombic and a tetragonal phase were included in the refinement. Similar to the quenched samples o f Jorgensen et al. (1990a), the slowly cooled samples of Radaelli et al. (1992) do not show the abrupt decrease of the c-axis found by Cava et al. (1990) (fig. 12) and shown by equilibrium samples (fig. 25) in single-phase Rietveld refinements. The dependences o f the lattice parameters on the oxygen content are shown in fig. 32. An inflection can be seen in the c-axis in the region around x ~ 0.35, similar to quenched samples (Jorgensen et al. 1990a). The change from the tetragonal to the orthorhombic model in the refinement has been done by determining the composition ( x - 0 . 3 ) o f the transition via extrapolation o f the orthorhombic strain to zero as shown in fig. 33. Allowing for partial occupancy of the oxygen sites around Cul and using isotropic temperature factors for all atoms, the peak widths were calculated with an isotropic strain parameter a 1, including the instrumental resolution and isotropic particle-size broadening. The chain (04) and antichain (05) site occupancies are shown

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30 x 1.0

,i~l,lI,

l,l~l,l,l,[

0.8

U

0 1 i III

0.6

III

o

0.4

E

c >. x

0.2

12) 0.0

61

I

6~ ,~o ° ,~ ,

0.0

0.2

0.4

O5 ~ ,o~,ag~,o,o~,

0.6

0.8

1.0

Fig. 34. Chain (O1) and anti-chain (05) site occupancies as a function of oxygen nonstoichiometry. Note the different notation of the chain site in this figure. The transition T - O takes place abruptly, in contrast to previous investigations (cf. fig. 13). After Radaelli et al. (1992).

in fig. 34. The antichain sites are occupied to ~5%, a fact not considered by Cava et al. (1990), which has appreciable influence on the BVS calculations (B6ttger et al. 1996). The occupancy of chain oxygen (O1 in Radaelli's nomenclature) shows a plateau-like deviation from linearity in the range x ~ 6.42-6.50. In the orthorhombic range there is no converging of the two curves but an abrupt change to tetragonal. This is opposite to the picture given by the high-temperature in situ measurements either at constant oxygen partial pressure (Jorgensen et al. 1987b) or at constant temperature (Shaked et al. 1989, Jorgensen et al. 1988). There, in the orthorhombic regime the 05 occupancy increases smoothly to match the O1 at the transition, a typical behavior of a second-order transition exhibited also by the samples quenched at high temperatures (fig. 13) (Jorgensen et al. 1990a). Also, the dependence on oxygen content of the oxygen-site order parameter [n(O1)-n(O5)]/[n(O1) +n(O5)], which is the order parameter of the transition, shows a strong discontinuity (fig. 33) which is compatible only with a first-order phase transition. Again this is in contrast to the earlier results of Jorgensen et al. (1990a), where the plot of the two geometrical order parameters of the T ~ O transition [orthorhombic strain b - a, and chain site occupancy n(O l) -n(O5)] is linear and extrapolates to zero, a clear criterion for "a well behaved 2nd order transition". The picture becomes clearer if the two order parameters, site occupancy and orthorhombic strain, are plotted one vs. the other, both for the slow-cooled samples (Radaelli et al. 1992) and for the in situ measured high-temperature (490°C) samples (Shaked et al. 1989, Jorgensen et al. 1988). Whereas the in situ data lie on a line through the origin, the line of the data from the slowly cooled Er-123-Ox samples has an intercept of 0.65! Thus, whereas the high-temperature in situ measurements display a second-order behavior, those o f the slowly cooled samples.follow a first-order behauior.

For a first-order transition we expect hysteresis resulting in a mixture of orthorhombic and tetragonal phases. Indeed, two-phase refinements by Radaelli et al. showed a larger c-lattice parameter for the tetragonal phase (fig. 32) leading to a discontinuity at the transition. Summing up, there are several arguments supporting the idea of a first-order T--+ 0 transition - at least at R T and lower temperatures (Radaelli et al. 1992). According

62

E. KALDIS

to the theoretical considerations of Kartha et al. 1995 this is probably a martensitic transformation. As mentioned above (sect. 3.2.2.3), the RT X-ray lattice parameters of the equilibrium samples indicate a higher degree of complexity with the different transition widths for the c- and a-, b-axes (figs. 25a,b,d). A comparison of all samples (XRD, NPD at RT and LT) discussed in this review shows that the lowest x value for the appearance of the T-O transitions in the a- and b-parameters is 6.3 for equilibrium samples and 6.35 for quenched and gettered samples. The highest value for its completion is 6.45. The X-ray RT c-parameter of the equilibrium samples (fig. 25d) shows a more abrupt step which seems to extend only in the 6.40-6.43 range (although some small deviations from linearity appear also up to 6.5). This behavior of the c-parameter seems consistent with a first-order structural transition. As already mentioned, structure refinements with synchrotron radiation may clarify the situation. The question has been asked many times whether the T - O and I-M transitions coincide, as originally suggested by Cava et al. 1990. The equilibrium samples are superconducting at x=6.396, but small amounts of superconducting material appear already at x = 6.342 (fig. 27b). For the gettered samples the onset of superconductivity was x = 6.45 and for the quenched samples 6.34 can be extrapolated (Jorgensen et al. 1990b). We may conclude, therefore, that the I-M transition takes place in the course of the structural T - O transition, which may possibly follow two steps: first intra-plane ordering and then adjustment of the intra-plane spacing. Thus, the investigation of equilibrium samples seems to support the original idea of Cava et al. (1990), that the onset of superconductivity is triggered by the contraction of the apical bond.

3.3.2. Raman inuestigations near the T - O transition Recently micro-Raman investigations (sect. 5.5, 6.2) could be extended over the whole range of insulating and superconducting 123-Ox (7 > x > 6) (Palles et al. 2000a, Liarokapis 2000, Liarokapis et al. 2000). As we will discuss later (sect. 5.5.1), the oxygen-site phonons show appreciable changes in the whole nonstoichiometric range. This is also true for the x < 6.5 range discussed here, which includes the I-M and T - O transitions. Figure 35a shows the dependence of the apex (Ag) phonon frequency on the oxygen nonstoichiometry, at RT. The figure includes data taken with three different excitation wavelengths. Strong (vertical) shifts appear in the range of the transitions (6.3 <x < 6.45) particularly between blue (476.2cm -1) and red (647.1 c m l ) , but also between green (514.5cm -1) and red excitation. This is due to resonance excitation of the 486cm l (x=6.2) and 493 cm 1 (x=6.3-6.4) modes as discussed in the past (Liarokapis 1997). The apex is abnormally wide, indicating the coexistence of several phases (Iliev et al. 1993, Liarokapis 1997). As will be discussed in sect. 5.5.1, a deconvolution with four Lorentzians (Palles et al. 2000a) shows the coexistence of four phases (475,486, 493 and 502 cm-1) (fig. 63), indicated in fig. 35a by the horizontal lines. We prefer (sect. 5.5.1) to show the individual data coming from the excitation of several parts of the same crystallite and also from different crystallites, in order to scout for the appearance of different phases in the same sample. Following the green excitation spectra (solid squares, fig. 35a) we find

63

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN Y B a 2 C u 3 0 ~

5,9 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 7,1

"7,

5

............... ;i ..... i ...... : ......... i ....... : . . . . . .

495~ ..............._i_.~ m~o..'~

505-

505

500-

500

495-

YBa2Cu316Ox in_ ~ k ~ Oapex(Ag) r ~ ±~

65 60

....................... 490 +" 490-i S~ ~ T~

50

'6

,~E 485: ~ i

480t

• •

: '.o ~ i i i'647.10m4

i i

Liarokapis

480

etal.

content

YBa2Cu316Ox O p l (Blg)

• 40

.

°~ 475-

475b .............. ;...; ..... i....,....,....,..;..,....,....,... 475 5,9 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 7,1 oxygen

i

~

~

.35 ~-

470•

(a)

x

•

465

1 35 (b)

c,

30

25

460 -20

4552

'1

'l

' F

I '1

'l

'l

'l

'l

'

6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 &9 7.0 o x y g e n c o n t e n t (x)

~- 340'6

35 25

e,

,~ 335-

"~

o= 2O

~ 30'-I

o

~25 e"' Q,,, ] ~20fl

330 -

_

•

4

I 'I

'I

' I 'I

'I

' I '

'I

10 4

"-

i

•

!i

i

i

, • :

•

• •

i :

:

•

•

.-"

•

• m~/~':n m

m

"'~m|

,.,,~"

"-

:1

"~li i " i • , , , =, : : :

...............

' I '

. :

? ,& : i

--..

m

6.0 6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9 7.0 o x y g e n c o n t e n t (x)

. : : :

: :~ : : ~ : : , nl

] |

-15 325 -

•

123-O, × DarokapJs

Raman et al.

-

: .....

i ..................................

5,9 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 7,1 oxygen

content

x

Fig. 35. (a) Changes of slope at x ~ 6.2, 6.3, and 6.4 of the apex phonon frequency vs. oxygen content, due to the N6el, T O and I-M transitions (see text). Solid squares, green laser; solid squares in a frame, blue laser; open circles, red laser. Note the shifts of frequency for the compositions x ~ 6.22, 6.28, and 6.4 and excitation by two different wavelengths, due to resonance. The horizontal dashe~dotted lines correspond to the frequencies of the phases deconvoluted (fig. 63) from the apex width. (b) Top: Apex phonon frequency vs. x, average values over many measurements. Bottom: Apex phonon width vs. oxygen content. Note the maximum and the minima associated with the above transitions, cf. (a). The large width of the apex phonon is an indication of the coexistence of several phases (see text). (c) Abrupt discontinuity in the Raman frequency spectra vs. x of the B lg phonon (out-of-phase vibrations of the 02 and 03) at x ~ 6.25 (top), coinciding with the onset of the Fano coupling with the superconducting carriers shown by the increase of the width vs. x of the B l g phonon (bottom). (d) In-phase (vibrations of the 02 and 03) phonon width vs. x. See text. After Liarokapis et al. (2000).

abrupt c h a n g e s o f s l o p e or d i s c o n t i n u i t i e s at x ~ 6.2, 6.3 a n d 6.4, i n d i c a t i n g the e x i s t e n c e o f three transitions, t w o o f t h e m o v e r l a p p i n g w i t h the a b o v e - d i s c u s s e d o n s e t o f the T - O a n d I - M transitions.

64

E. KALDIS

As mentioned above, in the small length scale of Raman spectroscopy (sect. 5.5.1) the 486 and 493 cm < phases coexist (phase separation) in the oxygen-content range of the T - O transition. Inspection of the apex phonon width (fig. 35b) shows a sharp maximum in the range o f x ~ 6.35, starting at x ~ 6.15, and followed by a sharp minimum at x ~ 6.45. Near x ~ 6.0 the apex frequency tends towards 475 cm -1 (fig. 35a) and, therefore, as seen from fig. 64, also a 475 cm -I phase is present there, decreasing at x ~ 6.2 and disappearing at 6.4. Also, the magnetic properties at room temperature change at this composition range. The N6el temperature is 4 1 0 K at x ~ 6.05 and 250K at x ~ 6.34 (Burlet et al. 2000). Linear extrapolation gives x ~ 6.25 for 300K. This is where the first transition takes place in fig. 35a. The oxygens of the planes seem also to become involved in this transition, as an abrupt decrease of the B 1g phonon (02, 03 out-of-phase vibrations along the c-axis) frequency takes place at x ~ 6.25 (fig. 35c, above). The width of this phonon also shows a sudden increase in the same region with an asymmetric Fano shape due to coupling with carriers resulting from the charge transfer (fig. 35c, below). Therefore, we reason that the onset of carriers in the planes appears at RT at the N6el temperature i.e. at x ~ 6.25. We may assume that these charges are the result of the formation of hole-doped clusters in a tetragonal insulating matrix, leading, with increasing doping, to the onset of the bulk T - O transition at x ,~ 6.30 (as seen by the X-rays) (figs. 25) and the I - M transition at x ~ 6.4 (figs. 27). The in-phase Ag phonon width (in-phase vibrations of 02, 03 along the c-axis) also shows an anomalous behavior in the range of these transitions (fig. 35d). Although the baseline shows that a minimum of the width should appear at 6.2 < x < 6.4 the scattering of the data indicates appreciable contribution due to cluster formation or disorder. In the tetragonal phase the same phonon shows appreciable anharmonicity (Liarokapis et al. 2000), disappearing at the T - O transition at x ~ 6.30 and increasing again at x > ~6.40. The localization of carriers in the insulating phase seems to influence the anharmonic potentials at x < 6.3. The anharmonicity disappears at the structural transition. But the delocalization at x > 6.40 supports the anharmonicity (possible polaron formation) which, however, decreases with increasing carrier concentration to disappear at optimal doping (Liarokapis et al. 2000). In conclusion, the above structural, optical and magnetic investigations suport the picture of three sequential transitions at RT correlated with the onset of superconductivity. The complexity is increased by the coexistence of three phases at mesoscopic scale and the possible appearance of hole-doped orthorhombic clusters in the tetragonal insulating matrix.

4. Local structure

Historically, the interest in the local structure has arisen from investigations of the nonperiodic structure of liquids and particularly of amorphous solid materials. The term "local" is often used to indicate that a view of the structure is given as seen from

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa~Cu30 x

65

the atoms inside the solid. Several methods give information about the local structure. Among them are the "extended X-ray absorption fine structure" analysis (EXAFS) and the "pair distribution function" (PDF) analysis of neutron powder diffraction data. Neither are restricted by the symmetry conditions of diffraction. Lattice imaging and electron diffraction techniques also give information about the local structure but are subject to the laws of diffraction. As an additional disadvantage of these methods may be considered the thickness of the sample (>200A) which is averaging the signal over this range. Other methods giving information about the local structure are NQR and M6ssbauer spectroscopy. Particularly for the former the body of literature is so large that its discussion would by far exceed the boundaries of this work. In the previous sections we have discussed some lattice effects which are to some degree reflected in the average structure (e.g., lattice-parameter anomalies). Local deviations from the average structure become apparent in crystallographic analysis with higher Debye-Waller factors. The Debye-Waller factor of the split atom model of the chain oxygen will be discussed in sect. 4.1, and in the following subsections we will discuss the results of other methods of investigating the local structure of 123-Ox. The large number of phonon investigations, e.g., with inelastic neutron scattering is outside the scope of this materials-oriented review. A recent exhaustive reference for this field exists (Furrer 1998). 4.1. Anomalies of the Debye-Waller factor: the split atom model of the chains Indications for local structure anomalies emerge from the value of the Debye-Waller factors (B) resulting from the average structure refinements. A good example in the case of the HTc superconductors is the chain oxygen deformation. According to the classical Debye theory for a harmonic crystal B i i = 82~U 2

with u 2. the mean square atomic displacements in the i direction, given in ~2. Already Capponi et al. (1987) observed, in the structure refinement of YBa2Cu3OT, much larger thermal vibrations of the chain oxygen 0 4 perpendicular to the chain (b-axis) and anomalous temperature dependence. Using anisotropic thermal vibrations in their refinement they found much better agreement. Figure 36 shows the temperature dependence of the B33 factor for the anomalous chain oxygen 0 4 compared with the normal behavior of the apical oxygen O1. In the same year these results were reproduced by two other groups (Beno et al. 1987, Beech et al. 1987). Soon, a much more accurate neutron diffraction study of oxygen-rich polycrystalline 123-O6.91 was performed (Francois et al. 1988). A higher precision by approximately a factor of 2 was achieved mainly by including higher-order reflections. A possible constant scale error in the neutron wavelength was reduced by normalizing the unit cell volume to that established by precise X-ray measurements. This study also found the anomalously high B factors of the chain oxygen 04. They could reduce them by using, in the refinement, a split-atom structural

66

E. KALDIS

::t, .Lo ; ............ . ~'~ 2,5

~ 2,0t

~

Capponietal.1987

1,°1

i

o,~1 ~ 0

::

50

100

o~/~-o l

I

150 200 T [K]

250

300

Fig. 36. Debye-Waller factors as a function of temperature. Note the much higher value and the different temperature dependence of B for the chain oxygen 04 as compared to the other oxygen atoms of the structure, e.g. B for the apical oxygen O1. After data of Cappuni et al. (1987).

model for the chain. This assumes a displacement of the 0 4 atoms into two potential minima perpendicular to the chain at a distance ~±0.1 ,~ along the a-axis. Thus, the chain is certainly not linear and has probably a (randomly) disordered zig-zag shape. The justification o f this model came from the resulting normal values and temperature dependence of the isotropic B factor. A detwinned single crystal of 123-O6.7 gave an X-ray refinement supporting the above findings (Wong-Ng et al. 1990). Later, an untwinned crystal of 123-O6.88 was precisely investigated and electron density maps could be constructed by Sullivan et al. (1993). They also confirmed the anomaly o f the 0 4 and suggested from the electrondensity difference maps that a dynamic disorder rather than a static split-atom site exists. Refinements with the split-atom model or with anisotropic 0 4 B factors gave results similar to those o f Francois et al. (1988). Oxygen-rich single crystals (x=6.98 and 6.96, T~=90-91 K) were also investigated by Schweiss et al. (1994) with high-resolution elastic neutron scattering. Their data confirm the work o f Francois et al. 0 9 8 8 ) and their split-atom model for 04. They considered this static model to be more acceptable than dynamic disorder. Probably the sample history was also important here, because Pyka et al. (1993) when performing inelastic neutron scattering in a good sample with x = 6.95 (To = 92 K, 95% single domain) discovered a new vibrational mode with frequency 5.1 THz = 21 mcV = 170 cm -I at F. In addition, its polarization and absence in the 123-O6 spectra qualify it for a transverse (100) 0 4 vibration. The ratio linewidth/frequency at 0.07 was rather large, and it did not decrease appreciably at low temperatures. The rather weak temperature dependence of this peak was considered not to be evidence for an instability o f the chain oxygen with its strongly temperature-dependent B factors. The increased value of the linewidth/frequency ratio can be explained either with moderate anharmonicity or with strong coupling of this phonon to electrons. Based on their elastic studies mentioned above the authors concluded that there is clear evidence for anharmonic behavior. Anharmonicity is an important issue in the debate about the mechanism o f superconductivity because it can lead to enhanced electron-lattice coupling. A brief discussion of this issue for the chains of 123 is given by Pickett (1995); the references include

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN Y B a 2 C u 3 0 ~

67

many articles on this subject. A volume edited by Mihailovic et al. (1995) discusses anharmonicity in HTcS. 4.2. Lattice distortions: investigations of the pair distribution function (PDF) A detailed discussion of the PDF analysis has been given by Egami (1994), in Kaldis (1994), and by Egami and Billinge (1996). The latter is a recommended review of lattice effects on HT~ superconductors up until 1996. The Pair Distribution Function (PDF) represents the weighted probability to find an atom at a given distance from another atom. The elastic scattering part of the diffracted intensity is the sum of the Bragg diffraction intensity and the diffi~se scattering intensity. The latter is very important for investigating the local structure because it contains information about the non-periodic components of the structure. The difference between the conventional crystallographic approach using the Rietveld refinement and the PDF becomes clear if we recall that the former uses only the Bragg intensity, leaving out the diffuse scattering. Thus, no aperiodic features are included in the crystallographic refinement so that an average structure results. On the contrary, the PDF analysis of the diffraction data renders the local structure. Another advantage of the PDF analysis is that it is not necessary to assume a certain structural model, although in fact variations of the parameters are often used to make comparisons between a calculated model and the PDE The scattered intensity can be measured as a function of the momentum transfer Q = k - k0 (k, scattered, k0, incident wave vectors). In measuring the PDF it is possible to reach very high Q values up to 25-30 ~-1 (d-spacing: 0.20-0.25 A), whereas using Cu K~ radiation only Q << 8 ~-1 can be reached. Rietveld refinements reach Q < 15 ~ 1 (d-spacing: 0.4 A). For the field of interest of this review, these investigations brought an important result:

local lattice distortions of the order of O.1A in domains of the order of lOA exist in HTc superconductors, which in contrast to some of the effects discussed in sects. 3.2.2 and 6.1 are not reflected in the average structures. This intrinsic inhomogeneity has nothing to do with doping defects, as it appears in chemically homogeneous materials like YBa2Cu408 (124) which does not have oxygen nonstoichiometry due to the stabilization of the double chains by edge sharing (Karpinski et al. 1988, Kaldis et al. 1991). It is also present in 123-Ox with x ~ 7.00 which certainly does not have gross nonstoichiometric deviations. For HTo superconductivity these effects are important in view of the similar dimensions of the coherence length, and as a possible indication for the existence of polarons. The interested reader can find a complete discussion in the review of Egami and Billinge (1996). The lattice distortions found by the PDF investigations of Egami and coworkers in various HTc superconductors are mainly triggered by displacements of the oxygen atoms (surrounding the Cu atoms) along the c-axis. However, for optimally doped 123 the results of the PDF investigations were rather ambiguous (Egami and Billinge 1996). Recently, two PDF investigations (Louca et al. 1999, Gutmann et al. 2000) have dent with the apical bond problem in 123, the latter as a function of stoichiometry; these will be discussed in sect. 4.4. These results are related to the high-resolution ED investigations

68

E. KALDIS

(Etheridge 1996) discussed in the next section, and the EXAFS measurements of R6hler et al. (1997b) which we will discuss in detail in sect. 7.4. 4.3. D&tortion of the planes." high-resolution electron diffraction Etheridge (1996) presented a systematic in-depth ED study of nearly stoichiometric 123. Although the investigation of other oxygen contents has not yet been published by this author, we include her results in this review, as they give information about the distortions of the superconductive planes. Several techniques have been used to achieve this result: (1) In selected-area ED patterns oery long exposures could discover a structured weak diffuse scattering (figs. 37 and 38). Static correlated displacements of atoms in directions nearly (2803) and (091) are proposed as the reason for this diffuse scattering. These are consistent with coupled displacements of Cu2 and the apical oxygen O1 sites in the a-b plane, which are nearest neighbors in these directions. The coupling occurs over a distance of two unit cells as illustrated in fig. 39. (2) In all high-resolution electron microscopy (HREM) images of 123 published up to now an extremely subtle modulation of the atomic contrast can be found along some zone axes, which remained unobserved in the past. The interpretation given by Etheridge (1996) is that it results from a local perturbation of the charge density distribution along planes parallel to {449}. The oxygen pyramidal planes defined by the plane oxygens 02, 03 and the apical oxygen O1, shown in fig. 5, are parallel to these planes. The dimensions of this perturbation along the (094) and (904) directions (roughly along the c-axis) are 5 2t, and along the (110) directions (in the superconducting plane) 10-20 A. Therefore, these are intersecting planes dividing the

Fig. 37, Twinned orthorhombic crystallite 123-O69, Selected-areaED pattern of the (111) zone axis. Diffuse streaks (along the arrows) parallel to (10[)* and (01 i)*, The intensity distribution of these diffuse streaks is characteristic rather for atomic displacementsthan for substitutions or vacancies. After Etheridge (1996).

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30 ~

69

(a)

Fig. 38. (a) High-resolution electron microscopic image along the (111} zone axis of 123-O x (with x > 6.9). The scale bar is 10A. (b) The corresponding Fourier transform shows diffuse scattering streaks linking Bragg reflections parallel to the (101)* and (011)* directions in reciprocal space. These appear also in the diffraction pattern of the same crystallite (fig. 37). After Etheridge (1996).

pyramids

square planar roups

Fig. 39. Schematic average structure of 123-O6.9 with atomic displacement vectors [2803] and [2805] (arrows). They illustrate that the coupled displaced atoms Cu2 and O1 (apical) ' a r e two unit cells apart. The (091} displacement vectors are directed into the paper plane and are not shown in this figure which is only two unit cells deep. After Etheridge (1996).

70

E. KALDIS

d~094~

d<094> is less than 5A d ranges from -10 to 20A

\ {449} cell walls

Fig. 40. Schematicillustrationof the network of irregular cells, with walls parallel to {449}, partitioningthe superconductingplane of 123-Q9. After Etheridge (1996).

superconducting planes in cells of dimensions comparable to the coherence length. The irregular network of these shallow dish cells is schematically shown in fig. 40. The walls of these cells are parallel to the oxygen pyramidal planes shown in the previous figure. In successive superconducting planes different patterns of cells may exist. (3) High-resolution dark-field images suggest that adjacent cells are slightly misaligned, inducing the atomic displacements. (4) Convergent-beam ED (CBED) patterns along the minor zone axes [01 l] and [101] show that the mirror planes perpendicular to the a- and b-axis are absent. This is further evidence for local symmetry breaking in the average Pmmm structure (in addition to the few oxygen vacancies, 1:25 sites). It is important to stress that the atomic displacements found by ED (1), are consistent with the local perturbation of the charge density found in the HREM images (2). Also, the range of interaction of the displacements from (1) is consistent with the frequency of appearance of the cell walls received from (2) (Etheridge 1996). Summing up, it can be said that the above phenomena are very feeble and need particular effort to be observed. However, it is gratifying to see that they all independently contribute complementary evidence for the model given above ~'g. 40): superconducting planes as a network of irregular cells 2-4 unit cells wide, resulting from correlated displacements of Cu2 and O1 two unit cells apart. Depending on the phase correlations of these displacements it is possible that a coupled tilting of linked arrays of the Cu05 pyramids results (Etheridge 1996). Etheridge stresses particularly the fact that the dark line contrast of the HREM images is present in most publications up to now and therefore it can be considered as an intrinsic property of 123-07. She also gives extended proof that the above effects have nothing to do with oxygen vacancies in the chains, irradiation damage or tweed structure phenomena. As the origin of the cell network of the superconducting planes again lattice strains, already found by the BVS method (compare sect. 3.2.1.4), have to be considered. From the coarse order of the cells, long-range strain fields can be assumed. This supports theoretical models discussing nanoscale patterning (e.g., Bishop 2000). According to Etheridge, several effects influencing the elastic energy can be responsible for these strains:

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTS 1N Y B a 2 C u 3 0 ~

71

(a) The different ideal dimensions between the Cu2-O2,O3 superconducting planes and the C u l - O 4 chains as well as Ba-O planes. It is reasonable to expect that the relaxation of the Cu2-O bonds in the superconducting plane will distort the apical site, as this is the most feeble point of the structure (Cu2-O1 is the longest bond, sect. 2 ) . (b) The orthorhombic strain due to the different lengths of the a- and b-parameters is responsible for the atomic displacements of the 0 4 chain sites found by Francois et al. (1988) (see sect. 4.2). These displacements may also influence the other bonds of Cul with oxygen shifting the position of the apical oxygen O1 (Etheridge 1996). We recall that steric reasons were given also by Brown (1989) sect. 3.2.1.4 for the strains in the 123-Ox lattice. Two completely different approaches are, therefore, leading to similar conclusions. In conclusion it can be said that electron diffraction shows complex oxygen displacements both perpendicular to (like Egami with PDF, sect. 4.2) and in the planes. The nanodomain structure of the superconducting planes found by Etheridge can be considered as a kind of crossed stripes system offerroelastic origin. It triggered investigations of its influence on macroscopic transport and magnetic properties of 123-Ox (Jung et al. 2000, Darhmaoui and Jung 1996, 1998). These authors found that the temperature dependence of the critical current density depends on the disorder of the superconducting planes. Further, they were able to determine the correlation between the superfluid density at low temperature and the nanoscopic disorder in the planes of 123 thin films. They also estimated that for thin superconducting films (thickness 1500A, area 1 cm 2) the abovementioned nanostructure (fig. 40) increases the surface area of the (110) face by almost six orders of magnitude. No wonder, therefore, that its influence is quite visible in macroscopic properties. Out of the wealth of the distortions investigated with inelastic neutron scattering (Furrer 1998) we chose arbitrarily to mention here the recent work of McQueeney et al. (1999) and Egami et al. (2000). Reinvestigating the anomalous dispersion of the longitudinal (LO) phonons (Pintschovius and Reichardt 1998) in La S r cuprates this group found a discontinuity in dispersion, indicating dynamic short-range cell doubling in the CuO2 planes along the direction of the Cu-O bond, probably a charge-ordering effect different from the stripe charge ordering. Egami et al. (2000) investigated the spatial charge inhomogeneities in manganites and cuprates. In manganites these are polarons, in cuprates the inhomogeneities have more dynamic nature and involve phonons that result in charge transfer between oxygen and copper.

4.3.1. Nanoscale phase separation. The Phillips model The evidence for the existence of nanoscopic disorder as an intrinsic property of 123-Ox brought by the above-mentioned work of Etheridge, is an appreciable support for the theoretical model introduced by Phillips (1987, 1990, 1991, 1999-2001) and by Phillips and Jung (2001a,b). A main feature of this model is that in contrast to all other theoretical models for HTSC it is not based on the effective medium approximation (EMA) but is

72

E. KALDIS

inspired by the glassy state, an idea which had been also discussed very early by Mtiller et al. (1987). According to Phillips, the conventional field-theoretical models derive the properties of the material substituting the intrinsic spatial inhomogeneities by the EMA. This is supposed to include the up to now unknown but essential characteristics of the material responsible for the anomalous properties. In contrast, the model of Phillips is based on the existence of intrinsic heterogeneities resulting from nanodomains in the HTCS. This is of course an approach with special appeal to those engaged in materials research. The importance of the nanodomains derives from breaking the CUO2 planes into a

system of metallic nanoislands and a network of semiconductive nanodomain walls that contain a pseudogap (Phillips 1987). Thus the electrical conduction through the planes shows a pseudogap confined to a single plane. Electrical conduction is also possible outside of the planes at frequencies below the gap, via resonant tunnelling dopant centers in the semiconducting planes (Ba-O) which separate the cuprate planes. Using such zigzag filamentary paths it was possible to provide new explanations for neutron diffraction anomalies (Phillips 2000, Reichardt et al. 1989), IR (Homes et al. 1995) vibronic spectra (Phillips and Jung 2001a), and STM (Phillips and Jung 2001b, Hudson et al. 1999, Pan et al. 2000, Howald et al. 2001). Recently, an increasing consensus has emerged about the existence of intrinsic inhomogeneities in HTC superconductors leading to phase separation (e.g. Howald et al. 2001). We note that elastic energy seems to play an important role in HTSC, associated with various lattice distortions (cf. sect. 3.2.1.4, Fig. 20; sect. 4.3, Fig. 40). The well-known formation of misfit dislocations in thick epitaxial layers (Frank and van der Merwe 1949, van der Merwe 1963) offers a good analogy to the formation of the nanodomains of the superconducting planes (Phillips and Jung 2001b). With increasing thickness of the epitaxial layer, the lattice constant misfit leads to increasing stress. As the harmonic strain misfit energy increases quadratically with the layer thickness, at a certain thickness the strain exceeds the elastic limit, and misfit dislocations appear in order to decrease the elastic energy of the system. Misfit dislocations in semiconductors appear commonly by shifting a lattice half-plane. However, in case of phase separation relief of the elastic misfit can be achieved by the existence of any second phase with different lattice constant. If the energy difference of the two phases is small, the reduction of the total energy can be achieved even by a single atomic layer. The nanoscalephase separation model of Phillips and Jung (2001b) assumes that this is the case for the cuprates with their easily distorted perovskite structures. 4.4. EXAFS and PDF investigations of the apical oxygen The Extended X-ray Absorption Fine Structure (EXAFS) is generated by the excitation of a certain atom in the unit cell, using X-ray photons with energy beyond that of the absorption edge. The oscillating absorption pattern from which the local structure is calculated, results from the interference of the emitted photoelectrons and those backscattered from the atomic cluster surrounding the absorbing atom. The strong scattering of electrons by atoms makes EXAFS an extremely sensitive probe of the local structure. The complexity of EXAFS spectra is the result not only of single scattering

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN Y B a 2 C u 3 0 ~

73

paths but also strong contributions from multiple scattering paths. The latter enrich the spectra strongly and if they can be analyzed offer a great number of structural details on the coordination sphere around the excited atom. In addition, in EXAFS the momentum transfer for the backscattered photoelectrons is very high, reaching values higher than 16A -1, and giving much more structural detail than the crystallographic Rietveld refinements. A resolution in momentum transfer of 0.1 A 1 in the radial distribution function of EXAFS in the range k = 4-14 ~-1 could be reached with Rietveld refinement only for Q ~ 25-28 ~-1 (R6hler 1994). Very short length and time scales strongly increase the importance of EXAFS investigations. The length scale is of the order of a few ,4 and the time scale is of" the order of <
74

E. KALDIS

o

-2 -4 I

I

)

I

I

I

I

I

l

I

I

1

I

I

I

I

I

I

I

I

i

i

i

,

,

,

~'--' 7 8

K I

I

I

i

i

,

i

I

0

-4

~. ,~,

i

i

I i

i,

*

,

I

,

i~l

i l l l l , l

2 0 -4

-, 2 / ' "~

,

,

,

t

,

- i ~4 i , i i i

LI

I ~ l l l l

4

I

p

I

,

I'"

i i I i i i I

6

,

,

I

I

l l l l l l l l l l l l ,

. . . . . .

:,,,,,,,

-4

,

8

Ill

10

,I

I

y

Iii

12

ook'

iIi

'

tl

14

k(l-') Fig. 41. EXAFS k~(k)oscillation spectra of YBa2Cu2.gssNi0.04206.95 depicting a beat near 12A a which was originally assumed to result from a splitting of the apex site: experimental (solid) curves at various temperatures mad calculated fits (dashed) assuming the split apex site. After Booth et al. (1996).

Billinge (1996) report an unpublished PDF investigation o f Billinge et al. where real space structure refinements do not show any disorder o f the apical oxygen site, in agreement with the crystallographic results. Nevertheless, i f in the refinement they use a split position o f only 0.033 A for Cu2 along the c-direction they arrive at smaller agreement factors, in accordance with the slightly elongated Cu2z thermal factor. They conclude that this is possibly an indication for inhomogeneous charge distribution in the planes, indicative o f the existence ofpolarons or a microscopic phase separation o f holes, in which case

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN Y B a 2 C u 3 0 x

75

a Cu2 atom would adjust its height along the c-axis when a localized hole was in the vicinity. These results are supported by the recent work of Gutmann et al. (2000). Booth et al. (1996) suggested that a comparison should be made between the correlated Debye-Waller factors given by EXAFS, which give the instantaneous positions of the atoms, and the uncorrelated average over many unit cells given by diffraction. To this end they deduced, from the instantaneous deviation of the atomic position from its mean position, the average deviation in the interatomic distance of atoms A and B, the broadening parameter OAB (Debye-Waller factor), and from that a "correlation" parameter q~: 02B = 02 + 02 -- 20"AOBG where O2ABis given by EXAFS, and o 2 is given by the diffraction experiments. The expected values for q} are +1 for in-phase vibration of atoms (acoustic phonons), -1 for out-of-phase (optical phonon) and 0 for atoms several cells apart above the Debye temperaturel Figure 42 shows the average deviation of the broadening parameters OAB as a function of temperature. Note that in the figures 0 4 labels the apical oxygen instead ofO1. The Debye-Waller factors increase with temperature as expected from a not high Debye temperature. An exception is the Cul-O1 bond which, due to an anomalous double peak in the region of T~, has a much higher average value in the temperature range considered. Also, the Cu2-O1 bond reflects the anomalous behavior of the apex oxygen near Tc with a clear peak. Both Stern et al. (1993) and Kimura et al. (1993) have found anomalies of the Cul-O1. Booth et al. (1996) conclude that although the physics behind these oscillations of OCul ol and OCu2-Ol cannot be yet finally explained, they can be interpreted as changes of the correlation of the Cu2, Cul and O 1 positions in the lattice. In agreement with this, the correlation parameter q) of the Cu2-O 1 pair shows clear anomalies with a stepwise decrease in the range 80-100 K (fig. 43). According to these authors it is possible that the increase of the Debye Waller factor and the decrease of the correlation parameter may result from the appearance of a negatively correlated mode. This mode excited near Tc would counteract the correlation dominating in the remaining temperature region. Such mode could be of polaronic-hopping nature (Booth et al. 1995). The dynamic biStability in the apex O1-Cul-O1 cluster was modeled by Mustre-de Leon et al. (1992) by assuming tunneling of the holes between two states leading to a double-well potential due to the large polarizability of the apical O 2 ion; this is generally compatible with current polaron models for electron-phonon coupling. The calculations of Ranninger and Thibblin (1992) showed that with decreasing speed of hopping the PDF becomes wider and then splits. A decrease by a factor of <2 is enough to bring about this change: When the frequency of hopping of the electron exceeds that of the lightest optical phonon, obviously the lattice cannot respond and the displacement of the individual atoms remains small; at slower hopping the lattice has time to respond and the Cu and O1 atoms come nearer or move away with the polaron hopping in and out of the ligand. If the hopping frequency becomes smaller than the optical frequency two

76

E. KALDIS 1.00 0.045 -

Cu(1)-O(4) 0.95

S

0.040

-

'

0.90. .z.

0.850.035 0.800.030

' ' ' ' 1 ' ' ' ' 1 ' ' ' ' 1 ' ' ' '

0

50

100

150

0.75

'

200

0

'''~1''''1''''1''''1' 50 100

T (K) 0.08

200

0.8

Cu(2)-o(4)

0.7 0.6

0.07

9E ~-

0.06

0.4

0.2-

....

I"'''1 .... 50 100

I .... I' 150 200

T (K) Fig. 42. The Debye-Waller factors a~B VS. T (see text) for the single scattering (SS) paths. Only the C u l - O 4 and Cu2-O4 pairs (here, 04: apex) show anomalies near To. After Booth et al. (1996).

t

0.5

0.3

0.05

150

T(K)

t ''°'l''''l''''l''''l' 50 100

150

tt 200

T(K) Fig. 43. Correlation parameters ~ vs T for the C u l - O 4 and C n l - O 4 pairs show clear anomalies near Tc (see text). 04 indicates the apex site here. After Booth et al. (1996).

different pair distributions appear (= polaron absent or present). Thus, a small change in the hopping rate of the polaron could explain the difference found between the single crystal (split atom) and the film (single position). Also, the wider Debye-Waller factor of the Cu2-O1 peak compared to Cul-O1 and its correlation parameter ~0.5 are consistent with the Ranninger model. Optical measurements (1993 state of the art) do not seem to support the existence of a double-well potential at the apex site. As discussed in controversy (Mustre-de Leon et al. 1993, Thomsen and Cardona 1993), it should correspond to an IR mode which could not be detected up to now with IR and Raman measurements. Further, the temperature dependences of the frequency, linewidth and oscillator strength of the IR mode of the

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS 1N YBa2Cu30 x

77

apical oxygen have step-like discontinuities at To, whereas the variations in the energy levels of the double well show a singularity. R6hler et al. (1992) have also found a step-like discontinuity near Tc in the Cu K edge EXAFS, supporting the shift of the chemical potential at the onset of superconductivity (Khomskii and Kusmartsev 1992). Analyzing the EXAFS beat structures of Mustrede Leon et al. (1992) and Stern et al. (1993), R6hler (1994) presented a different interpretation, which is in agreement with the finding of Stern et al. that the beats disappear in the oxygen-rich range. His model calculations confirmed the existence of two C u l - O 4 bond lengths with a difference of ~0.1 A in the EXAFS spectra, but offered an alternative explanation by a static model based on the local distortions of Cu2 caused by isolated oxygen vacancies of the chain. Nevertheless, as pointed out by Egami and Billinge, the discrepancy with the crystallographic Debye Waller factors remains. The calculation gives a characteristic life time of the two states of 10-13 s. In view of the EXAFS time scale of ~10 15 s and its ability to give instantaneous pictures of the local structure, R6hler (1994) argued that no difference is expected between a two-site apex configuration resulting from dynamic bistability and one resulting from static vacancyinduced distortions. However, the instantaneous picture delivered by EXAFS also seems to be subject to yet unknown boundary conditions. This is illustrated by the valencefluctuation phenomenon of some rare-earth ions in their solid compounds (R6hler 1994). They can be described as dynamic alloys of two different valence states of the same atom, occupying crystallographically indistinguishable sites with a fluctuation time scale of the order of 10 -13 S. T h u s , time scale and splitting of the bond lengths of the atoms in the two valence states are comparable to the double-well apex configuration discussed above. EXAFS spectra show beats for the "inhomogeneous" mixed valence systems where the valences are static or slowly fluctuating (~10 9 s), but they show almost no effect in "homogeneously" mixed valence systems fluctuating with ~10 13 s. The reason is not known. Kohn et al. (1982) proposed that the unstable wave functions hybridize with their nearest environment. As mentioned above, very recently two new PDF investigations on the same subject have appeared which we will briefly discuss here. In a reduced-dimensional function like the PDF a complex crystal structure like that of 123-Ox may lead to a situation where several atom pairs have overlapping bond-distance correlations. The angular averaging then leads to the loss of all information about orientation. The result is that the C u - O pair correlations and the associated local distortions can not be resolved from those of other pairs like Y-O, Ba-O, and O-O. To overcome this difficulty Louca et al. (1999) applied the isotope Difference Pair Density Function (DPDF) analysis in two 123-O6.92 samples made of the pure isotopes 65Cu and 63Cu. Due to the difference in the neutron scattering length of these isotopes, and the resulting enhancement of the scattering amplitude, this method is a Cu-specific PDF analogue to the EXAFS measurements at the Cu absorption edge, but possibly with higher resolution. The local atomic structure around Cu and its temperature dependence is obtained without interference from the undesirable pair correlations of Ba-O, Y - O etc. which have comparable bond distances.

78

E. KALDIS

Louca et al. (1999) have worked with seemingly well-characterized samples (no quenching, annealing at 400°C for 48 h), although they do not disclose lattice parameters and some other structural data. They found significant structural distortions in the superconducting planes, associated, however, mainly with the Cu2 and not with the apical oxygen. They also found a temperature-dependent splitting of the Cu2-O1 pair correlations. However, they could locate its origin to the Cu2 sites because, if it originated from the apical oxygen, then similar anomalies should appear in the C u l - O 1 correlations, and this was not the case. The splitting is found to be of the order of 0.25 A, but due to technical reasons (limitation of the Fourier truncation) the authors do not consider this value reliable. They conclude that the double Cu2-O1 peak does exist, with 2 of the pairs close to 2.25 A separation and 5 with a somewhat larger value. The authors point out that the nature of the distortion of the superconducting planes they find is consistent with the previous findings of R6hler et al. (1997a,b) about dimpling (Cu2-O2,O3) which will be discussed in detail in sect. 6. The direction of the Cu2 distortions is, therefore, parallel to the c-axis. At the time the manuscript of this review was nearly completed the PDF work of Gutmann et al. (2000) appeared. As the authors note a disadvantage of their investigation compared with EXAFS is the measurement of the total PDF (disadvantages discussed above) and not of the chemical specific PDE Gutmann et al. (2000) investigated the PDF of four different stoichiometries, x = 6.25, 6.45, 6.65, 6,94. Unfortunately, half of the samples were investigated in another neutron facility so that there is no quantitative agreement between the two groups of data. The authors did not find any evidence for the splitting of the apical oxygen site. However, they arrived at a slightly improved fit if the Cu2 site was split along the direction of the c-axis. This is interpreted as the result of charge inhomogeneities in the superconducting planes. They also discussed possible formation of stripes (see sect. 7.4). We note here that a splitting of the apical oxygen signal is routinely seen (Haase 2000) with a novel NMR technique in 123-O6.95 (Haase et al. 1998). At present it is not clear whether it is a static or a dynamic effect. Recently two different apex signals have been found also in La2 xSrxCuO4. The above discussion shows the challenges which HTc materials are issueing to all scientific disciplines involved. In view of all the difficulties discussed above, it is necessary to expand the investigations to a much wider range of oxygen contents. The present knowledge supports the existence of anomalies first observed by the Los Alamos group. After a series of investigations it is found that the source of anomalies is not the apical oxygen but the Cu2 site distortions which seem to be important for HTc superconductivity (e.g., Andersen et al. 1995). As the work of Booth et al. (1996) indicated, the Cu-O1 pair distribution function is more complex than the originally assumed split-atom model. A concluding remark for this section concerns the quality of the samples. In several cases sample characterization data are missing (e.g., values of lattice parameters, methods of synthesis, magnetic data), giving the unfortunate impression that when so sophisticated investigations are involved the quality of the samples does not matter!

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS 1N YBa2Cu30~

79

5. Insulating, underdoped, optimally doped phases: phase diagram, phase separation and superstructures 5.1. Phase diagram and superstructures." experiments and calculations More information about the superstructures will probably improve our understanding of the charge transfer mechanism. An open question is how the electronic states of the chains hybridize with those of the planes. Both the local, intra-chain oxygen order controlling the length and disorder of the chain, and the inter-chain order leading to the superstructure formation are not well known. The same is true for the relationship of oxygen disorder and the superconducting properties. As a result, only idealized models can be investigated up to now. From the experimental point of view the difficulties are very large due to the natural limitation of the freezing mobility of oxygen.

5.1.1. Experimental investigations." EM, XRD and ND The dependence of the critical temperature on the nonstoichiometric superstructures resulting from oxygen ordering of the chains was found very early. To our knowledge, the existence of the Tc ~ 60 K phase was first reported by Tarascon et al. (1987a) and the first two-plateau curve by Johnston et al. (1987) and Cava et al. (1987a). The shape of this curve (for slow-cooled equilibrium samples, cf. fig. 2) triggered a very large number of EM investigations which brought to light the existence of a series of superstructures. These and several other investigations (see, e.g., in Kaldis 1990) showed at an early stage the importance of the EM (HREM and ED) investigations not only to discover superstructures, but also to give a first framework of a structural model for the XRD and NPD investigations of new structures of HTc and related compounds. Van Tendeloo et al. (1987) discovered that, in addition to the existence of the full-chain, normal-123 structure, a0, a 2a0 superstructure (every second chain empty) appeared for x > 6.5 if oxygen was stripped off the chains via electron-beam heating. More systematic investigations were then performed as a function of nonstoichiometry by Chaillout et al. (1988), extended later to larger series of nonstoichiometric samples (AlarioFranco et al. 1988) and followed by the discovery of the herringbone superstructures 2 x / ~ ° 2 x / ~ ° c type (at x ~ 6.75 and 6.25) which recently was reinvestigated and found to be due to impurities (Yakhou et al. 1996). Among the large number of other early papers dealing with EM investigations of superstructures we mention Zandbergen et al. (1987), Werder et al. (1988), Reyes-Gasga et al. (1989) and Beyers et al. (1989). The investigation by Beyers et al. (1989) is particularly important because the authors investigated equilibrated samples, synthesized under well-controlled conditions in a solidstate ionic cell (Ahn et al. 1988), using slow cooling (18°C/h). This is in contrast to most other investigations at the time, which were performed on quenched, nonequilibrated samples. The result was increased homogeneity. Thus, with the exception of the x = 6.65 and 6.90 samples, all investigated crystallites in samples of the same stoichiometry gave reproducible electron diffraction patterns (Beyers et al. 1989). These

80

E. KALDIS

authors found long-range ordering along the c-direction only for the 2a0 superstructure (doubling of the a-axis, wavevector (½,0,0) which appeared in the x=6.28-6.61 range covering both insulator and superconductor fields. A two-phase region was found for x=6.65 with the 2a0 and an ordered superstructure with wave vector (2,0,0). Due to the lack of investigation of the intermediate compositions, we can assume that this twophase region covered the range 6.61-6.71 (fig. 109). A sequence of higher superstructures with short-range ordering was found in the range x=6.71-6.90. At x=6.71 with wave vector (0.37,0,0); at x = 6.75 with wave vector (½,0,0); and at x = 6.85 and also in some crystallites of the the x=6.90 diffuse streaks along the a-direction. Again it seems reasonable to assume that the diffuse streaks extended in the range 6.85-6.90 (fig. 109). Beyers et al. (1989) discuss their results in the frame of the Gibbs phase rule, tending to the classical picture that the plateaus of Tc are two-phase regions (miscibility gaps) and the rapidly changing Tc regions x ~ 6.40 and 6.80 are homogeneity ranges, i.e single phases (solid solutions with oxygen). This was illustrated (Beyers and Shaw 1989) by a schematic figure similar to fig. 45 but without the lattice parameters. Based on this thermodynamic picture the Tc, onset results of fig. 2 may be interpreted as showing a sequence of superstructures in the "60 K plateau". But, there is more evidence from fig. 2 that a miscibility gap (two-phase region) appears in the x ~ 6.75-6.80 range. The lattice parameters - a more direct criterion for the coexistence of phases - indicate also that a series of two-phase regions appear in the superconducting range (sects. 3.2.2.3, 3.2.2.4). This picture is also supported by the results of Raman investigations (sects. 5.5.1, 5.5.2). More details are given further below and in the discussion of the important points of the T-x phase diagram (sect. 9). The superstructures with wave vectors ~2 at x ~ 6.65 and 0.37 at 6.71 found by Beyers et al. (1989) were not identified. The diffuse streaks at x ~ 6.85 indicate a series of very similar superstructures, which were considered by de Fontaine et al. (1990a) as the result of their structure combination branching mechanism (sect. 5.2). They also considered the 0.37 as a 3 which according to their ASYNNNI model could give a chain sequence (11011010) of full (1) and empty (0) chains. As in EM-investigations only small parts of the sample volume can be investigated, and nonstoichiometric artifacts are easily possible due to overheating of the material by the electron beam, very early, bulk structural techniques like XRD and ND were used to prove the existence of superstructures. The 2a0 superstructure was observed for the first time with XRD by Flemming et al. (1988) and the 3a0 by Plakhty et al. (1992). Further important work with XRD was performed by Zeiske et al. (1992), Grybos et al. (1994), Plakhty et al. (1994), Schleger et al. (1995a,b), Poulsen et al. (1996), Andersen et al. (1996), E. Straube et al. (1998), and very recently with hard X-rays by von Zimmermaun et al. (1999) as a function of stoichiometry, which we will discuss in detail. The first observation of the 2a0 with ND was reported by Zeiske et al. (1991) and of the 3a0 by Plakhty et al. (1995) and Schleger, et al. (1995a). One of the important contributions of this XRD and ND work was to show - in contrast to the early ideas, and consistent with local-structure work (sect. 4) - that in addition to the oxygen ordering of the C u l O chains appreciable relaxation of the cations is associated with the formation of

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN YBa2Cu30x

81

Table 9 Antiparallel atomic displacements in the 2% (ortho-II) and 3ao (ortho-III) superstructures of 123-Ox" Atom

Meanposition in unit cell ao bo Co

Atomic displacements (]~) Ortho-II 2aox

Y

½

Ba

½

Cu(1) Cu(2) 0(1) 0(2) 0(3) 0(4)

1 1 1 12 1 1

±2 !2 0 0 0 0 ½ 1

2± z 0 Z Z Z Z 0

0.011(1) 0.039(1) 0 0 0 0.002(1) 0 0

Ortho-III CoZ

0 0 0 0.015(2) 0.040(2) 0 0.004(1) 0

3aoX

0.010(1) 0.059(2) 0.034(2) 0.007(1) 0.014(1) -0.006(4) 0.007(5) -0.045(6)

CoZ

0 0.040(2) 0 0.009(1) 0.068(8) 0.005(7) 0.010(7) 0

a After Plakhty et al. (1995).

these superstructures. Table 9, after Plakhty et al. (1995), shows the ant±parallel atomic displacements in the 2a0 (ortho-II) and 3a0 (ortho-III) superstructures as determined for both phases by joint profile refinement o f X-ray and neutron scattering data, measured at RT, on single crystals. Plakhty et al. (1994) have found the 2a0 superstructure in crystals with 6.40 ~>x >~ 6.63 and the 3a0 superstructure with 6.70 >~x ~> 6.80. In the range 6.63 ~ x ~> 6.70 they found a two-phase region. In spite o f the different error margins o f the determinations o f oxygen content (Beyers et al., Ax = -4- 0.02, Plakhty et al. neutron refinement, Ax ,,~ -4-0.4) and particularly the different thermal and chemical history o f the samples, there is a partial overlap o f the miscibility gaps. The miscibility gap (2a0 +3a0) o f Plakhty et al. overlapps with the two-phase region o f Beyers et al. (1989) and the range o f the ortho-VIII superstructure (consisting o f a sequence o f 2a0 and 3a0 chains) found by von Zimmermann et al. (1999), (cf. fig. 109). Their results (Plakhty et al. 1994) show for a very slowly cooled crystal (x=6.55) only short-range order even for the 2a0 superstructure, with maximal anisotropic correlation lengths 240(30)A along the b-axis, 17 A along the a-axis [100] and 8 . 6 A along the c-axis [001]. Similar domain sizes have been found before by Hohlwein (1994). Pinning o f the domain walls by impurities, twinning and other defects, as well as the high activation energy for the motion o f the C u - O chains, are the reasons considered. Figure 44a shows the structures and the shifts o f the various atoms in these superstructures. The oxygen ordering results in appreciable relaxations o f the cations which can be seen clearly in the average crystallographic structure. The largest impact has the shift o f the Ba atom. Shifts which appear in both structures include: (a) Ba 2+ shift towards the full chains. The x-displacement appears stronger in the 3a0 superstructure which has two full chains. The z-displacements in this structure are

82

E. KALDIS

Y Z

~

Cu(2) !

o0) 0

c°,,,/!

0

U r

1

(3

O(0 ( Cu(l)

O(l)

i

Cu(2) i

Cu(2)

¢o

Y

Ortho-II

•

I,

Y

Ortho-III

Fig. 44a. The average crystallographic structures of ortho-II and ortho-III phases. The arrows indicate the displacements of the various atoms due to the formation of the superstructures. After Plakhty et al. (1995).

larger for the Ba2 (of the central cell), which is under the influence of full chains from both sides. This results in a shift of 0.008(4) A towards the basal plane (b) O1 and Cu2 shift towards the empty chains in 2a0. In analogy it can be assumed that a similar shift of 03 appears in 3a0. Thus, Cu2 is displaced out of the plane already as a result of the formation of superstructures. This is interesting in conjuction with the PDF investigations discussed in sect. 4.2. (C) y3+ shift towards the empty chains. This is due partly to the shift of 03 and partly to the shift of Ba 2+. It is believed that these shifts are not significantly influenced by the oxygen content (Hohlwein 1994, Plakhty et al. 1995) and the variation of temperature (Straube et al. 1998). In the most recent single-crystal diffraction work with hard (100 KeV, penetration depth 1 mm) X-rays (yon Zimmermann et al. 1999), it was found that the only true equilibrium phases with long-range order are the T (only phase appearing for x < 6.35) and the O a0 (x > 6.82 at RT). The 2a0 is also a 3D phase, but with small domains - as mentioned above - appearing for 6.35 ~<x < 6.62. The 3a0 is a 2D single phase appearing in this investigation in the range 6.72 ~<x ~<6.82. In addition to the 2a0 and 3a0 superstructures, they found the following superstructures or mixtures of superstructures: (1) Ortho-V: a chain sequence (10110) resulting from the sequential ordering of 2a0 + 3a0. It is a bulk 2D phase and appears in a mixture with 2a0 at a composition x = 6.62. This is not far away from g3 (x = 6.6) which would be the ideal 5a0 composition. We recall that Beyers et al. (1989) observed at x=6.65 (6.61-6.71) with EM the mixture of 2a0 and 2 (x = 6.4); (2) Ortho-VIII: a chain sequence (10110110) resulting from the sequential ordering of ortho-III and ortho-V [3a0 +(2a0 +3a0)], with ideal compositions ~ (x=6.375)

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa~Cu30 x

83

1000 /

800

0 // /

T

D/

v

"~ I--

/

OI

/

600

\

i el/

400

200

it

0.3

,

,

,

i

0.5

i

L

I

0.7

,

,

.

0.9

x in YB~CusOe. x Fig. 44b. Superstructures of 123-O~ in thc T - x field, as determined by hard X-ray diffraction. The linger pm'abolic curve (dotted line) shows the prediction of the ASYNNNI model. The experimentally determined curve lies 200 K lower. After von Zimmermann et al. (1999).

and ~ (x=6.25). The actual modulation vectors at 0.382 and 0.627 are near to these values, the phase appearing for slowly cooled crystals at x = 6.67. It is a bulk 2D phase. Beyers et al. (1989) found the phase (0.37,0,0) at x=6.71. We note that 6.67 is approximately the range where the 4a0 superstructure could be expected. This superstructure has been reported only once as found by EM (Van Tendeloo and Amelinckx 1990) but was absent in the investigations of von Zimmermann et al. (1999). The appearance of 2a0 and 3a0 sequences before the pure 3a0 extends the contributions of 2a0 up to x = 6.72 and makes difficult the investigation of the properties of the pure superstructures. From the point of view of phase stability, it seems reasonable that the pure 2a0 and 3a0 phases are separated by mixtures with increasing 3a0 content. We recall that Plakhty et al. (1994, 1995) have found in approximately the ortho-VIII range [3a0 +(2a0 +3a0)] the two-phase region of 2a0 and 3a0 (cf. fig. 109a). Better crystals, even slower cooled and much longer than one-hour measurements may show even more clearly the boundaries between two-phase and single-phase regions. A very good account of the growth history and characterization of the crystals used in this study is given by yon Zimmermann et al. (1999). Using gas-volumetric equipment they can determine the oxygen content with a relative accuracy Ax = 0.02 using the ideal gas law, provided that the volume of the apparatus and the original composition of the buffer powder are exactly known. The accuracy of the method depends on this calibration. Figure 44b summarizes the phase relationships they have found. This yet incomplete T - x phase diagram is dominated by the tetragonal and the "ideal" (all chains occupied, intra-chain occupation statistical) a0 orthorhombic structure, which appears at higher temperatures (above T ~ 350K) down to x > 6.35. It is clear, therefore, that in this range

84

E. KALDIS

a large number of oxygen vacancies exists in all the chains of the a0 structure. According to the authors only these two phases are true equilibrium phases and show 3D long-range order. At room temperature the a0 orthorhombic structure appears only at x > 6.82. Characteristic for the phase diagram at x < 6.82 are phase transitions above room temperature (< 150°C) between the a0 orthorhombic structure and the superstructures. Transitions occur down to 30-50°C and it is possible to expect at least a tendency for such transitions even at lower temperatures. This feature accounts for many of the numerous contradictions and irreproducibilities in sample preparation and several property measurements of the last 15 years! Depending on the conditions of synthesis the activation energies for the appearance of the various superstructures will be different. The a0 ~ 2a0 transition takes place upon cooling at 368 K. This is 200 K lower than predicted by the ASYNNNI model (sect. 5.2; see also fig. 44b). As mentioned above, the single-phase, 3a0 superstructure has a finite-size ordering in the a-b plane. It does not appear in this work at its nominal composition x = 6.66 but in the range x = 6.72-6.82, with maximum intensity at x ~ 6.76. The a0 ---+3a0 transition takes place at 321 K. For both 2a0 and 3a0 superstructures the peak intensity and width freeze at T <35C °. The thermal behavior of the ortho-V and ortho-VIII phases is interesting. Heating of ortho-V leads to 2a0, but ortho-V (2a0 +3a0) is not yet recovered on cooling after one hour. The disappearance of the ortho-V correlations upon heating takes place in the range 50-70°C with the 2a0 correlations increasing simultaneously. The latter disappear near 110°C. Heating of ortho-VIII above room temperature leads to ortho-V and then to 3a0. Upon cooling, ortho-V is recovered in one hour. However, by longer duration of the experiments it is possible that 3a0 could also be recovered. The ortho-VIII superstructure peaks start broadening upon heating at T = 42-45°C. Above 50°C the peak position shifts to the position of ortho-V, and at 150°C has reached the position of 3a0. Upon cooling the scheme is reversible down to 75°C and then freezes into ortho-V (cf. fig. 44b). Characteristic of these XRD results (von Zimmermann et al. 1999) and of several previous findings - particularly of the EM investigations of Beyers et al. (1989) - is the fact that the superstructures appear at oxygen stoichiometries higher than their ideal compositions, the deviations increasing with increasing number of chains. This means that "empty" chains are partly occupied Von Zimmermann et al. (1999) estimate from their results an occupation of the empty chains of ~ 10% for the 2a0 superstructure and 30% for the 3a0 superstructure. The resulting disorder probably influences the Cu charges. Von Zimmermann et al. (1999) discuss the possibility that this shift of compositions is due to a temperature-dependent tilting of the phase boundary lines. Possibly then, at low temperatures the compositions of the superstructures could be near the ideal values. However, due to the very slow oxygen ordering kinetics, already near room temperature when the superstructures become stable during cooling, the phase diagram is frozen. The freezing temperature Jbr the chain movements was estimated to be 313K, and that for the individual oxygen jumps to be 250K. This latter temperature was estimated from the timedependent increase of Tc for quenched samples at these temperatures (cf. sect. 3.2.1.5).

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN YBa2Cu30x

85

i

lattice 3aram.

i -

magnet.I

re-

resist. :~

/

/

06.5

06.5+0xiI O,

Ox+ 071 07

Oxygen Superstructures

i

Fig. 45. Schematic to illustTate that the two plateaus of the T~ vs. x curve of 123-Ox could be explained by a classical picture of miscibility gaps. One superstructure is sufficient to explain the two plateaus. Based partly on a figure of Beyers and Shaw (1989).

The above investigations also touch the problem of the chemical phase separation. Beyers et al. (1989) interpreted the coexistence o f the 2a0 and (~,0,0) superstructures at x ~ 6.65 as a phase separation leading to the T ¢ = 6 0 K plateau. This idea was partly inspired by the early theoretical work o f Khachaturyan and Morris (1988) and Khachaturyan et al. (1988), which alone at that time could account for the superstructures observed. These authors considered the superstructures as metastable transients that precede the spinodal decomposition into the main stable phases tetragonal at x = 6.0 and orthorhombic a0 at x = 7.0. We believe that in view o f the complexity o f this system, the possibility of miscibility gaps does exist. As already mentioned, Beyers and Shaw (1989) discussed a decomposition between the superstructure phases x = 6.5 and Ok and the stoichiometric structure x = 7.0, and considered the To changes in the possibly existing miscibility gaps between these phases. Figure 45 (based partly on a figure o f Beyers and Shaw 1989) shows the consequences for the change in lattice parameter and in To as a function o f the oxygen nonstoichiometry. As expected from the phase rule, in the two-phase regions o f the miscibility gaps the properties o f the coexisting phases are fixed because the oxygen content is fixed (pseudobinary system YBa2Cu3Ox-O2). The figure shows that phase separation due to two miscibility gaps would be sufficient to explain the existence o f the two "plateaus" at Tc = 60 and 90 K. As mentioned above, the investigation o f equilibrium samples (lattice parameters) seem to be consistent with the Gibbs phase rule picture. Von Zimmermann et al. (1999) discuss kinetic reasons for the explanation of the coexistence o f 2a0 and ortho-V at x = 6.62. Based on their studies o f oxygen ordering kinetics (Schleger et al. 1995a,b, Kfill et al. 1999), and the lower transition temperatures (95°C and 50°C), they point out that the formation kinetics o f ortho-V are much slower than those o f 2a0. Thus, the coexistence o f 2a0 with ortho-V could be the result o f freezing mobility before equilibrium is reached.

86

E. KALDIS

The phase diagram of fig. 44b does not comply with the phase rule in its present form. Comparison with phase diagrams of the type of fig. 1 makes this even clearer. Although both diagrams cover the subsolidus field, the study of the hydride phases can be extended down to liquid-nitrogen temperatures due to the much higher mobility of hydrogen. Most obvious difference between the two diagrams is the absence of twophase regions (miscibility gaps) or phase-transition lines separating the superstructure ranges in the 123 diagram. Von Zimmermann et al. (1999) point out themselves that their phase diagram is not yet complete. Their conclusion is that the superstructures they studied are not equilibrium phases and it is possible that in more perfect crystals and with longer annealing times a more complex spectrum of superstructures may appear. We may add that much more insight could be gained by investigating a much larger number of compositions. Study of the trends could then contribute appreciably to a better understanding of the phase diagram. Thus, although the diagram of fig. 44b is the result of the most advanced diffraction study (up to 1999) it possibly gives only some of the phases involved and not their complete relationships. After discussing the changes of structural and some physical properties as a function of oxygen nonstoichiometry we will combine, in sect. 9, all the experimental results we have discussed in this review in a tentative T-x phase diagram. Associated with the problem of the equilibration is that of finite-range order. This is particularly important for the 2a0 superstructure with its 3D short-range order and the largest stability region. Schleger et al. (1995a,b) have suggested that this short-range order is due to the formation of anti-phase boundaries which limit the growth of the 2a0 domains. Impurity defects stabilize the anti-phase boundaries and slow down the moving of long Cu-O chains limiting the domain growth. The results ofvon Zimmermann et al. (1999) and Kfill et al. (1999) support these ideas. The question arises here whether these phase boundaries are also associated with the stripes which recently have been found to exist also in 123-Ox (sect. 7.4). Anyway, the short-range order is consistent with the nanoscale phase separation model (sect. 4.3.1). The small 2D domains of the short-range superstructures 3a0, ortho-V and ortho-VIII may also suggest that a random faulting sequence 2ao and 3ao underlies the ordering mechanism. This simple model has been suggested a long time ago by Khachaturyan and Morris (1990) who have calculated structure factors qualitatively similar to those measured now by von Zirmnermann et al. (1999). Von Zinunermann et al. (1999), however, believe that this finite order is due to the sluggish kinetics of oxygen at low temperatures. The ortho-V and ortho-VIII superstructures are formed only by very slow cooling, indicating that long-range interactions become strong only at low temperatures. Model simulations (Jensen et al. 1997, Monster et al. 1999) indicate also that long-range interactions are necessary for the superstructures to form. Thus, according to the authors, the slow oxygen ordering kinetics for moving long chains at low temperatures are the reason for the small domains of these large crystallographic cells. The phase diagram of fig. 44b, in its present form, leads to following conclusion. The 60 K "plateau" (x ~ 6.6-6.75, fig. 2) of Tc corresponds to the stability range not of the

OXYGEN NONSTOICHIOMETRY A N D LATTICE EFFECTS IN YBa2Cu30~

87

single-phase 2a0 superstructure, but to a series of single phases and mixtures of phases whose structural units are the 2a0 and 3a0 sequences: 2ao (x ~ 6.6) • 2ao + ortho-V (x ~ 6.65) • ortho-VIII (x ~ 6.70) ° 3ao. It is a matter of interpretation whether one follows the Gibbs picture with miscibility gaps or the ASYNNNI model based on the CVM method and discussed in sect. 5.2. The first was followed by Beyers and Shaw (1989), and the results of the lattice constants of equilibrium samples seem to support it. The ASYNNNI model is nearer to the ideas of von Zimmermann et al. Much more work (with many more stoichiometric compositions) is necessary to fred out whether there are miscibility gaps between ortho-V, ortho-VllI and 3a0. Also, the single-phase character of the ortho-V and ortho-VIII phases should possibly be better supported. The transition between the two plateaus appears in the boundary of the phases 3a0 and a0. The question whether this is a miscibility gap or a transition between two phases with different chain sequences cannot be answered with certainty at the present stage of knowledge. However, the important fact remains that in this range (x ~ 6.75-6.80) many physical and structural properties change, indicating a correlation between superstructure ordering and superconductivity. This multiphase transition region between plateaus is consistent (a) with the T~ measurements of equilibrium samples (fig. 2), indicating a mixture of phases (hatched areas), and (b) with the horizontal parts of the lattice parameters vs. x curves of the BAO samples (figs. 28 and 29). That the a-lattice parameter shows several such plateaus is not surprising in view of the slow kinetics and the extreme sensitivity to the chemical and thermal history. A detailed discussion of the changes in structural and physical properties appearing at x ~ 6.75-6.80 will be given in sect. 5.4.2. 5.2. Model calculations in view of the complexity of the pseudobinary YBa2Cu3Ox-O2 phase diagram ,the task for the theory is extremely difficult. In the optimism of the first hour, simple lattice gas models were proposed soon after the discovery of high-temperature superconductivity. They were based on the assumptions that the system can be described as a lattice of oxygen atoms only interacting with either long- or short-range potentials (independent of oxygen content and temperature). The "equilibrium" states may be obtained either by calculations with the mean-field approximation or by Monte Carlo simulations. Rather unexpectedly, this approach did help to some degree in predicting phase-diagram properties. Improved versions of the ASYNNNI model helped to increase our knowledge about superstructures and ordering mechanisms important for the electronic structure and the charge transfer. A detailed discussion of these models is outside the scope of this review. We discuss only some of the highlights of the ASYNNNI model which are of interest for experimental phase-diagram investigations. They illustrate the great complexity of the superstructure picture.

88

E. K A L D I S

The first two models followed quite different approaches. Khachaturyan and Morris (1988) calculated the T - x phase diagram in the mean-field approximation using longrange interaction potentials of the oxygens in the basal a - b plane defined by the chains (Cul/O4/O5). The long-range interactions include strain effects and describe the mesoscopic ordering (twin formation). As mentioned above, Khachaturyan and Morris (1990) suggested a simple mechanism of formation of the superstructures which produced structure factors to some degree similar to those recently measured. These ideas led, however, also to the conclusion that the diffuse scattering is the result of interstitial defects in the b - c plane, which has not been verified experimentally up to now. Von Zimmermann et al. (1999) point out that the observed oxygen ordering has 2D character as it results from the defects of the a - b basal plane CulOx, and that it has domain sizes much smaller than the twin domains. The other early model (ASYNNNI, de Fontaine et al. 1987), followed the opposite approach, using short-range effective pair interactions; despite its simplicity it succeeded in accounting correctly for the formation of the chains, the appearance of the tetragonal and orthorhombic phases, and the locus of their transition in the T - x phase diagram. It also succeeded in predicting superstructures but - as we have seen above (sect. 5.1.1) it gave a quite erroneous critical temperature for the ortho-II phase. The effective p a i r interactions V , are deduced from the general effective cluster interactions (Sanchez et al. 1984) using the cluster variation m e t h o d (CVM) approximation. Thus for the basal plane of 123-Ox (Cul-O, mirror plane of the structure) the effective O - O pair interactions are given by the linear combination of the total energies:

Vn =

i l w (n) 4k"OO

w (n) _ w (n) _ w (n) "~

+ "[211]

"OI2

"DO,,

(3)

with O the occupied oxygen sites and [] the empty oxygen sites, W~)o the total energy average over all configurations of [] and O having two occupied O sites in nth-neighbor pair position, and similar definitions for other W(n)'s. The energies W have values on the order of the cohesive energy of the materials and can be calculated by quantummechanical methods. The effective pair interactions V, however, are small differences of large numbers and orders of magnitudes smaller than the W's (de Fontaine et al. 1992). In the work of de Fontaine et al. the oxygen configurational statistics are modeled using a 2D Ising (lattice gas) model of the basal plane considering three interpenetrating square lattices (fig. 46a): (a) of the Cul atoms (assumed to be always fully occupied), (b) of the 0 4 and c) of the 05 oxygen atoms. In the tetragonal phase the latter two sublattices are occupied randomly, and in the orthorhombic stoichiometric material (x = 7) one (04) is fully occupied and one (05) is assumed to be empty. The model considers only the O-O asymmetric interactions up to next-nearest neighbors (NNN). This is denoted also by the name: Asymmetric Next-Nearest Neighbor Ising model (ASYNNNI). As mentioned above, the interaction potentials are considered to be independent of T and x. Three effective pair interactions (not pair potentials) were considered in the original model, all inside the basal a - b plane divided into two sublattices (04 and 05): V1 couples the two oxygen sublattices of 0 4 and 05 and is the inter-sublattice nearestneighbor (NN) interaction. The experimental fact that C u l - O 4 chains are formed in the -

OXYGEN

NONSTOICHIOMETRY

VI:NN intersublattice,repulsive V2: NNN intrachain,attractive O

A N D L A T T I C E E F F E C T S 1N Y B a 2 C u 3 0 x

89

V3 : NNN interchain,repulsive VI>V3>0>V2 O

V4 e-

tl

I

v2 •

=-

=-

l lv(I

I

.o

Cul

s I

•

0

•

1 o

.E

0

•

©

•

123-basal plane a - Axis

~- ~ r30

=,

05 sitessublattice : mainlyunoccupied 04 sitessublattice : occupied

Fig. 46a. Schematic illustration of the three sublattices Cul (solid squares), 04 (gray circles) and 05 (open circles) and the interaction potentials V used in the ASYNNNI model. For a discussion see the text.

structure shows that this interaction is a strong Coulomb repulsion, i.e. clearly positive. The other two interactions are intra-sublattice N N N less strong interactions. - V2 is the intra-chain (NNN) interaction between two 0 4 atoms mediated by a Cul atom. It is an attractive covalent interaction which together with Vl force the 0 4 on their chain sites and keep the 0 5 sites nearly empty at x >6.35 and lower temperatures. - V3 is a weaker Coulomb repulsive inter-chain (NNN) interaction which stabilizes the 2a0 superstructure. The anisotropy o f the model is introduced with the inequality V2 ~ V3, which also from a phenomenological point of view is rather clear. To account for the existing higher-order superstructures a series o f interactions along the a-axis V3 =J1,J2,J3,... (corresponding to higher neighbors) have been introduced (e.g., V4 in fig. 46a). Stability and ground-state analysis (de Fontaine et al. 1987, 1990a,b, Inoue et al. 1987, Wille et al. 1988) indicated the relative magnitude and sign of these effective interactions to be

VI > V3 > O > V2.

(4)

This inequality expresses the fact that only the 4-fold square planar or the 2-fold apical C u l - O coordination are favorable in the frame of the model [fig. 46b, (a) and (c)]. Other theoretical calculations have shown that the 3-fold coordination [fig. 46b, (b)] has higher energy (Burdett and Kulkarni 1989) and decreases the hole count in the planes and Tc (Ceder et al. 1991b). We note that, nevertheless, this coordination appears at the ends o f chain fragments. Ignoring the intra-chain disorder is one o f the drawbacks o f this model. First-principle total energy calculations (Sterne and Wille 1989) gave absolute values for the effective pair interactions V 1 / k B = 4278 K, V2 / kB - 1488 K, and V3 / kB = 682 K. Using their data, Ceder et al. (1990) were able to propose a phase diagram with a real temperature scale in K. The boundaries of this diagram which were in agreement with the experimental results are the T ---+O transition and its temperature dependence. Figure 47a

90

E. KALDIS C

C ,L

4-Cu

3-Cu

2-Cu

(a)

(b)

(c)

C

b

C;

a

Fig. 46b. C u l - o x y g e n coordinations discussed by the A S Y N N N I model. For a discussion see the text. After de Fontaine et al. (1992).

shows the T - x phase diagram of de Fontaine et al. (1990a) derived without fitting any parameter, and the experimental values measured by Andersen et al. (1990). However, the same group (von Zimmermann et al. 1999) recently showed (fig. 44) that the hightemperature boundary of the OII phase lies 200°C lower than theoretically predicted. This indicates on the one hand the appreciable achievements, and on the other hand the limitations of this model and the degree of its usefulness for experimental research. De Fontaine et al. (1990a,b) derived a family of superstructures with a structure combination branching mechanism. They used the linear chain Ising model to investigate the quasi-lD interchain ordering. Assuming lack of intrachain modulations and disorder, but fully formed (identical) chains, they investigated the repulsive interactions along the a-axis direction beyond V3. These long-range interactions follow a convexity relation and can be obtained by a structure combination branching mechanism (de Fontaine 1984). The ground states of these quasi-lD superstructures, stabilized by convex interactions, are established by stepwise combination of the ortho-I and ortho-II structures (Level I) (fig. 47b). Thus, combination of full chains <1) and every second chain full <10) results in <110) or ortho-II structure, the combination <10) and <110) results in <11010), that of <1) and <110) in <1110) or ortho-III structures, etc. At Level IV, four new superstructures are predicted with sequence of chains <1101010) (approximately in the concentration range 6.5-6.6), <11011010) (approximately in 6.6-6.65), <1110110) (approximately in 6.68-6.74), <11110) (approximately in 6.78-6.98), and <1) above x = 6.92. Higher-level superstructures are expected to appear as the temperature decreases. As the long-range interactions become less and less strong with the decreasing oxygen mobility at lower temperatures, random mixtures of the 1D structures of the previous level should appear. We note that that these idealized "full-empty" chain superstructures considering only 2D O - O interactions have not been observed up to now, with the exception of the ortho-Ii (2a0) and ortho-III (3a0). The above superstructures are considered to be modulations of the same phase, rather than separate phases. Therefore, according to this model, no two-phase regions in the Gibbs sense are expected at the predicted boundaries. The CVM approximation predicted the T-OII and OI-OII transitions to be first-order. The two-phase regions are shown by

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30x

91

X in Y B a 2 C u 3 0 x 1500

1.0

6.2 I

6.4

6.6

N

6.8

e. •

•

7.1

. ~ . • ~ . e. •. o ~ ,

o o o o o

o•

e ee•

. . . . . . .

•

~ .

1000 T

T[K]

~

OI

I

00

II

0.1

I

0.2

I

0.3

I

0.4

0.5

Oxygen concentration on basal plane Fig. 47a. Pseudo-binary T ~ phase diagram of YBa2Cu30 x calculated with the CVM approximation. The small solid circles are the experimental data of Andersen et al. (1990). The inserts are schematic illustrations of the 2D superstructures of the basal plane obtained fi'om Monte Carlo simulations. Large solid circles, oxygen; small solid cu'cles, Cul (large and small solid circles form the black chains); open circles, vacant sites. After de Fontaine et al. (1992). The measurements of the phase boundary of the T - O transition are in good agreement with the them'y. However, comparison with fig. 45 shows that the calculated critical temperature of the ortho-II phase (~570K) is much higher than the value recently found [~370K, von Zimmermann et al. (1999)].

the tie lines in fig. 47a. However, transfer matrix calculations and Monte Carlo simulations do not show any first-order transitions (Bartelt et al. 1989, Aukrust et al. 1990). A useful review about the ASYNNI model is given by de Fontaine et al. (1990a, 1992).

92

E. KALDIS

a <1 O>

0=0.25

~1

Level 1

C=0.5

c=0.33 <110>

<11010>

<130> oo0375

°-o.3o ~.

c;

'

"

co

Ill

¢o

"

iv

~j V

I

V

I I

I

0.25

0.30

0.35

0.40

0.45

(x--6.5)

(6.6)

(6.7)

(6.8)

(6.9)

I. 0.50 (7.0)

Concentration

0~

<1:DO>

c=0.30

0.00

z=6.60

<11010>

I

i

]

I/3

2,/5 3/5

<110(10)2> C=0'29 = Z=6"57 [ t

<(l |O)zlo> [ C=0"31 Z=6"63 I

3174/7

3/IB J/8

]

C=0"3751[" z=1'75

zrJ

114 I/2 314

C=0"37 ] [Z=6"71 [ .

2/7

5/7

I

I ~,=0.40 <1(0;=6.80 / [ [ I ]

1/5 2/5 315 4/5

Fig. 47b. (a) Branching algorithm for quasi-lD superstructures of the basal plane of YBa2Cu30x. The symbols 1 and 0 denote oxygen-full and -empty chains. The concentration scale x is given together with the basal plane coverage. (b) Diffraction patterns (Fourier amplitude-squared) along (h00) of the superstructures existing on Level IV (levels I --+V, decreasing temperature; z, oxygen content x). Superstructures appear at x = 6.57, 6.60, 6.63, 6.67, 6.71, 6.75, 6.80. After de Fontaine et al. (1990b).

A l t h o u g h the A S Y N N N I m o d e l reproduces excellently the temperature and c o m p o sition dependence o f the O - T transition and the existence o f the O-II phase, as we have seen it is 200 K w r o n g in the prediction o f the O-I to O-II transition temperature. Further, it does not account for the appearance o f the up to n o w observed higher superstructures (O-V a n d O-VIII, sect. 5.1.1) and it predicts always long-range order

O X Y G E N N O N S T O I C H I O M E T R Y A N D L A T T I C E E F F E C T S IN Y B a 2 C u 3 0 x

93

of the superstructures, instead of the short-range order found. In view of the extreme complexity of the 123-Ox superstructures it is clear that one cannot expect much more from such simple models. An intrinsic drawback of the model is that it is based only on the order of the oxygen sublattice. However, as shown by structural investigation, appreciable cation relaxations seem to be necessary for the stabilization of the superstructures (cf. table 9, fig. 44a). A series of publications have tried to compensate the various drawbacks of the ASYNNNI model. Andersen and coworkers have experimented with several extensions of the model (cf. references in von Zimmermann et al. 1999). Schleger et al. (1994) introduced the charge and spin degrees of freedom of the itinerant carriers and considered the entropy contributions resulting from them. The agreement with experiment is very good. In conclusion it can be said that, even with the above-mentioned drawbacks, the ASYNNNI model has considerably helped to understand the difficulties of the superstructure problem. The delicate balance resulting from eqs. (3) and (4) indicates the strong dependence on preparation conditions to be expected. The complexity of superstructures dictates extremely slow cooling and low-temperature annealing. Here, however, the freezing of the mobility of oxygen puts a natural limitation. Another interesting topic is the structure combination branching mechanism. As the relaxation of the cationic lattice and the disorder of the chains is not taken into consideration even if the method is thermodynamically sound, the predictions can be only be correct to a certain degree. The appearance of ortho-V and ortho-VIII in the structure as found by von Zimmermaml et al. (1999) with different chain sequences than those predicted above show that simple "telescoping" of the oxygen is not enough. Thus, in our opinion the question about the existence of miscibility gaps separating the various superstructures still remains open. Independently of the partial validity of the branching mechanism of de Fontaine, with decreasing differences of the energies of the various superstructures the system comes near to a random distribution of chains. In view of the small driving force and the low mobility of oxygen, the experimental proof of the miscibility gaps becomes then very difficult. On the other hand, the relaxation of the cationic sublattice may increase the involved interactions so that miscibility gaps can be seen,

5.3. Tc dependence on nonstoichiometry and hole concentration: applications o f the B VS method - II

The strong dependence of the temperature of the diamagnetic transition on the carrier concentration was one of the first subjects to be investigated after the discovery of HTc superconductivity. Shafer and Penney (1990), who were among the first to investigate this subject, found a parabolic dependence of Tc on the hole concentration and wrote the first review on the subject. The parabolic dependence was further supported by Tokura et al. (1988), Torrance et al. (1988) and Torrance and Metzger (1989). Later Whangbo and Torrardi (1990) showed that this applies to all p-type superconductors. Tokura et al. investigated a large number of substituted 123 compounds of the general

94

E. KALDIS 2.0 2,1 2,2 , ~ ~ '

1,9 n

,

2,3 2,4 2,5 , . , . ,

120 -

120 • A

De L e e u w et al.

1990

100~'

2,6

80-

o

•

100 80

•

60

E ~

i

4020-

: 4-coord. 5-coord. 6-coord.

•

• 40 o O

20

n

1,9

'

a

'

2,0 BVS

r

'

2,1

n

2,2 of CU2

'

n

2,3

'

i

2,4

(incI.O1)

'

n

2,5

0 2,6

Fig. 48. Correlationbetween the maximum T~...... of 19 superconductors and the bond valence sum of the Cu2 atom. This correlation exists if not only 02 and 03 but also the apical oxygen O1 are included in the calculation. After de Leeuw et al. (1990).

formulae Y(LaxBal-x)2Cu3Oy and (YxCa~-x)Ba2Oy each with a different total charge o f non-copper cations Q. Knowing that the degree of oxidation, which they put equivalent to the average charge p per bond, is very important for these materials, Torrance et al. (1988) measured it approximately by iodometric titration. From Q and p they could estimate indirectly the oxygen content y from the simple elect~roneutrality condition Q + 3(2 + p ) = 2y,

(5)

and they could show that the changes o f Tc found by variation o f the oxygen content (Engler et al. 1987, Tarascon et al. 1987b, Johnston et al. 1987, Cava et al. 1987a) are due to changes o f the charge in the [Cu-O] p+ bonds. Studies o f a series of well-characterized samples as a function o f nonstoichiometry showed the existence o f a maximum o f Tc at ~x = 6.92 (Graf et al. 1990, Rusiecki et al. 1990). Also, at this composition range the anomaly o f the c-axis (fig. 28a) was found by Rusiecki et al. (1990), which was the first evidence for the existence o f different materials properties in the then scarcely investigated ooerdoped range. Claus et al. (1992a) made the first Tc vs. x investigation o f the overdoped region with estimated x values, and measured a few compositions with lower To. Just before the manuscript was finalized a paper containing a brief summary o f the rather large literature concerning Tc vs. x or hole concentration (Schlachter et al. 1999) came to the attention o f the author. The interested reader can find in its introduction references in addition to those discussed in this review. Very popular among crystallographers and materials scientists for the estimation o f the hole density is the empirical BVS method (sect. 3.2.1.4). Starting from the work o f Brown (1989), de Leeuw et al. (1990) considered separately the bond valences o f copper and oxygen in the planes. They demonstrated a correlation between the maximum Tc, max and the in-plane bond valence sum o f Cu2 and o f Op for 19 superconductors, whose structures were known at the time (fig. 48). In the Vcu2 they included the oxygen contributions both from the Op (02, 03) and apical oxygens (O1). The correlation disappears if the apical oxygens are not considered. The work o f de Leeuw et al. demonstrated clearly that the

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS 1N YBa2Cu30x

95

valences calculated with BVS are relevant to superconductivity and proposed a method to calculate the hole densities o f the planes. Figure 48 shows that Vcu2 decreases with increasing To,max. Extending the work o f de Leeuw et al., Tallon (1990) proposed to use, instead o f Vcu2 or V0, their sum and difference as independent variables, due to the fact that the holes have a hybrid copper-oxygen character: p .~ V

=

(Vcu

2 -

2) + (2

-

Vo2) q (2

V+ = (2 - go2 ) -t- (2 - V03) - -

(VCu 2 -

-

Vo3 ) =

2 + Vcu2- V02 - V03,

2) = 6 - V~.2 - Vo2 - V03,

(6) (7)

with Vcu2 - 2 the excess charge on Cu2, and 2 - Vop the excess charge on each oxygen of 0 2 and 03. V is a parameter for the change o f hole density in the planes and it becomes zero for Cu 2+ and 02-. Tallon (1990) argued that an advantage o f V_, compared to the valence sum o f copper Vcu2, is its relative insensitivity to internal stresses as the effects on copper and oxygen would tend to cancel each other. V is mainly sensitive to shifts of the oxygens O 1, 02, 0 3 out-of-plane, as the in-plane bond sums cancel each other. V+ is a parameter for the shift o f holes from Cu to O along the C u - O bonds in the planes (distribution of charge between Cu and O). Contrary to V_ it is sensitive to strain, the effects on Cu and O adding to each other. This is meaningful only outside the range o f the charge shift very near to oxygen (localization, semiconductivity) or very near to copper. Plotting Tc vs. V_ results in the well-known parabolic curve in agreement with the earlier results o f Shafer et al. (1988), Sharer and Penney (1990), Tokura et al. (1988), Torrance et al. (1988), and Groen et al. (1990) with a maximum near ~0.17 holes/Cu atom. Based on the data for Laz_xSrxCuO4 and the Bi family, Presland et al. (1991) described the parabola with

r~

- 1 - 82.6(p - 0.16) 2,

(8)

which was considered as a generalized empirical equation of the approximately parabolic dependence of Tc on the number o f holes p per Cu2 atom. They claimed that it applies for all the p-type superconductors, the more recent studies converging to a maximum at p ~ 0.16. However, it is difficult to incorporate 123 in this picture because there is no direct method for the measurement o f p in 123 as a function o f the nonstoichiometry x, due to the existence o f two different Cu valences. To overcome this difficulty Tallon et al. (1995) had an interesting idea, which, however, had a doubtful experimental confirmation. In Y1 xCaxBazCu306 with all oxygen removed from the chains (cf. fig. 46b, part c) the Cul atoms can be considered to be in the Cu 1~ valence state. In this case we have for the planes p ~x/2, i.e. if V_ for the undoped 123-O6 becomes zero, only the holes from the Ca doping still exist (the question not asked at the time is whether these holes contribute

96

E. KALDIS 0,00 1,2

0,15

O,lO

0,05

0,20

i

0,25

0,30

i

i

0,35 1,2 1,0

1,0"

0,8"

0,8

[] +

[] Do

I""°E"0,60,4" 0,2" 0,0 o 0,00

0,6

o

o

0,4 0,2

+zx r~ = q

0,05

i

i

i

F

0,10

0,15

0,20

0,25

I,

i

0,30

0,0 0,35

Hole concentration p, or V-

Fig.49. Genericparabolafor normalizedTovs hole concentrationbased on BVS calculations,for LaI xSrxCu204 and Y1 xCaxBa2Cu307 0 after Tallonet al. (1995). The validityfor the Y-compoundis based on the assumption that 7 - 6 = 6.0 samples, completelystripped of their chain oxygen,are still superconducting.As shownin fig. 51 this is doubtful. Open circles: Yt xCaxBa2Cu306 from p=x/2. Open squares: from p= V in YBa2Cu307 with different 6. Solid diamonds:fromp = V ofY I xCaxBa.2Cu307_~with 6 ~ 0.04 and differentx. + symbols: fromp= V of Y1 xCa~Ba2Cu307 ~ with x=0.1 and different 6. Solid squares: La2_~Sr,CuO4. Open triangles: La2_ySr~CaCu206. to superconductivity). In this way, if the exact stoichiometry is known also the hole density will be known. What these authors overlooked was that not the hole density, but the mobile hole density is relevant here. As we will discuss below, this is supported by recent results. According to Presland et al. (1991), Cay-123-O6 and Y-123-Ox follow the generic parabola ofeq. (8) shown in fig. 49. Using t h e p =x/2 a n d / ) = V relations the dependence of the hole concentration on doping for both 123-Ox and Ca(10%)-123-Ox was calculated. The dependence shown in fig. 50 should be a rather smooth function with a change of slope at x ~ 6.4 and a saturation at x < 6.3 for Ca-123-Ox. In spite o f the opinion o f the authors, the curve for x = 0 (0% Ca) is not smooth and shows clearly that a discontinuity appears. The authors argue that even if they use for the calculation o f p = V_ the data of Cava et al. (1990), which as we have seen (sect. 3.2.2) show the abrupt step of Cu-BVS near the T --+ O transition, the curve remains smooth without an abrupt change near x = 6.5. Tallon et al. (1995) and Tallon (1990) conclude that this is an indication that, independently of the existence o f an oxygen superstructure or of substitutional Ca doping, the process of excitation for the underdoped and overdoped regime remains the same. The above results of Tallon et al. concerning the Ca-doping are questionable both from the point of view of the quality of the samples and the physics of Ca dopings. According to R6hler et al. (1999b), at least for fully oxygenated Ca-123-O6.95, it is not the Ca introducing additional carriers, but the dilution o f the Y-123-O~ which leads to the decrease of Tc (sect. 8.4)! Further, as Merz et al. (1998) have shown, Y1 xCaxBa2eu306

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30x 0,2g,0

6,4

6,2,

0,15-

o

o

0,10-

•

o .......... o oo . . . . . . . . . "1i""" .11 . . . . . . . 0,00- .....• ......

-0,05 6,0

.

,•

~'

30.

#-

...'

•

Tallon et al. 1995 616 content

2 4 6 8 1012141618202224 .,.,.,.,. .,.,., .,.,.,. y (16%) £) .5%)

B(Sttgeret al. 1994 x = 6.30 ( % ) : Meissner

.

/

..'

"

0,15 0,10

123

614

0,20

..!'"

II I '"''

u

6,2

Oxygen

40.

O,25

o

o

Q. 0,05-

50

7,0

,

Ca(10%)-123 o

E

6,8

,

o

0,20,~

6,6

,

97

618 x

0,05 0,00 -0,05 7,0

Fig. 50. Bond valence sum parameter V vs oxygen deficiency 6 for Y~ xCa~Ba~Cu307 ,5with Ca contents x = 0 and x = 0.1. After Tallon et al. (1995). The x=0 curve does show a discontinuity (see text).

50

40

0

:::

30

2

20 ¸ = .0 (<4.2)etalWK. / 10. N?2~Tc I:/:

~'- Tallonet al. 1995 X = 6.0

2 4 6 8 1012141618202224 Ca%

2010

Fig. 51. Tc vs. Ca doping for Yj yCayBa2Cu306 fi'om data of Tallon et al. (1995) and Merz et al. (1998) together with some unpublished 06.3 data of B6ttger and Kaldis (1994). All Ca compositions of Merz et al. are not superconducting. The 06.3 samples of B6ttger and Kaldis show low Meissner effect, indicating that some of them are not really superconducting.

is not superconducting for most relevant Ca concentrations (0, 3, 10, 12, 23%). Figure 51 shows the Tc's found as a function o f the Ca concentration for Y1 xCaxBa2Cu306 by Tallon et al. (1995) and Merz et al. (1998), together with unpublished results o f B6ttger and Kaldis (1994) for three 06.3 samples with exactly determined oxygen content (sect. 3.1.3). We see that even at x = 6.30 some samples show marginal superconductivity with v e r y s m a l l Meissner effect. Several problems arise from the preparation o f such samples. Thus, Tallon found difficulties because Ca substituted both the Y- and the Ba-site in his polycrystalline samples. Oxygen-rich, equilibrium, polycrystalline samples do not have this problem, as NPD (B6ttger et al. 1996) and C a - E X A F S (R6hler et al. 1999a) have clearly shown. In single crystals, however, this problem does appear (B6ttger et al. 1997, Schlachter et al. 1999). Tallon et al. (1995) used high temperatures for the synthesis (up to 1000°C) in pure oxygen and cooled slowly. The Ca content was known to the authors only from,

98

E. KALDIS

unfortunately unpublished, neutron diffraction data. The oxygen content was known for most samples fi'om the weight changes (probably Ax ~ 4-0.2-0.4?), and only in a few cases from iodometric titration. Probably, the samples of Tallon et al. did have oxygen contents >6.0 and/or very small Meissner effect. But also from the BVS point of view several arguments have appeared recently against the concept of Tallon (eqs. 6, 7). The assuption was made by Tallon that the BVS directly gives the actual charge on the ion (Chmaissem et al. 2001). This approach differs from that of Brown (1989, 1991) who did not use both the BVS of Cu2 and that of Op, but only that of Cu2, after making corrections for internal strains. Brown used the fraction of Cu 3+ in the Cu2 plane and considered them to give the hole concentration in the planes. Chmaissem et al. (2001) compared these two different calculations for 123-Ox and (CaxLal_x)(BaL75 xLa0.25+x)Cu3Oy. Following Brown (1989, 1991) they use the fraction ~ of Cu 3+ in the Cu2 plane, the fraction of the Cu 2+ ions being (1 - ~ ) . The average charge (oxidation state)/Cu ion is: Vavg = 3~ + 2(1 - ~) = 2 + ~.

(9a)

This average charge should be equal to the sum of the bond valences of Cu 2+, --17(2+) = l:i~ ' /q+2~__OCu , and of Cu 3+, V(3+)= BVScu,+3 Vavg = ~ V (3+) -1- (1 - ~ ) g (2+).

(9b)

From eqs. (9a) and (9b) we arrive at V (2+) - 2

= V(2+) + 1 - V(3+)"

(10)

Similarly a ~(1) fraction can be defined for the Cu ÷~ in the chains ~(1) =

V (2+) - 2 V (2+) q-

1 - V (1+)"

(11)

Chmaissem et al. (2001) found that the calculations of the BVS in the past (Tallon 1990, Cava et al. 1990, Brttger et al. 1996, Karppinen and Yamauchi 1999) have not considered the presence of Cu +1 ( ~ = ~ 0 ) = 0 ) . Eckstein and Kuper (1999) have recently used eqs. (10-11) to calculate from the data of Cava et al. (1990) the V for optimally doped 123. Tallon (1990) and Tallon and Cooper (1993) have calculated with = 0, p ~ V ~ 0.16 and Karppinen and Yamauchi (1999) calculated V_ = 0.99. Eckstein and Kuper (1999) found for optimally doped 123 V_~0.105, showing that there is no universal value. According to them the discrepancies arise from the inconsistent use of ~=0. Chmaissem et al. (2001) calculated from the data of Cava et al. (1990) for oxygen-rich 123 the hole concentration C of the Cu2 plane using the method of Tallon with ~ = 0, the method of Tallon with ff ~ 0 and the method of Brown. The results are

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30x

99

Table 9a Calculation of the p and C values by Chmaissem et al. (2001), using the BVS method of Brown (1989, 1991), Tallon (1990) and Tallon and Cooper (1993), based on the experimental data of Cava et al. (1990)~

6.95 Tc pBL'ow. p~--OTallon p~=0 Tallon CB..... (%) C~-° Tallon (%) C~ ° Tallon (%)

90 0.264 0.105 0.165 53.9 23.0 36.7

Oxygen concentration x 6.84 6.81 88 0.238 0.089 0.150 58.8 23.3 44.1

86 0.222 0.075 0.135 69.8 26.1 43.4

6.78 80 0.207 0.068 0.125 71.3 24.3 44.6

After Chmaissem et al. (2001). For a discussion see the text. shown in table 9a, and should be compared with the experimental findings o f Zhu and Tafto (1997) who have used a novel electron diffraction method and found close to optimal doping C = 76-t-8%. The results o f the method used by Brown are the highest and nearest to the experimental results. The differences o f the method used by Tallon when using = 0 and ~ ~ 0 are also clear. In view o f all the results discussed above the validity of a universal parabolic

dependence of Tcvs hole concentration including 123 (fig. 48) is questionable. 5.4. Pressure dependence of Tc as a Jknction of nonstoichiometry: giant dTc/dP effects"

a t x ~ 6 . 7 5 andx~6.4 Pressure increases the charge transfer between planes and chains because it reduces their distance and, therefore, changes the hole concentrations. This effect is analogous to the change by variation o f nonstoichiometry. Thus, variation o f pressure gives an additional degree o f freedom to study lattice effects driven by changes o f nonstoichiometry. One o f the first groups who recognized and exploited the pressure effects in HTcS was that o f C.W. Chu (Wu et al. 1987). F r o m the strong To increase with pressure they had measured for the La compound o f Bednorz and Mfiller (1986) they got the inspiration for the discovery o f YBazCu307 x which strongly enhanced the HTc research. In principle, one would expect in the complicated perovskite structure o f 123-Ox pressure induced phase transitions, due to the large number o f structural degrees o f freedom or at least atomic rearrangements under pressure, shedding some light in the problem o f superstructures. Indeed, pressure investigations do illuminate - among other properties - the peculiarities o f the structure o f HT~S and to some degree its influence on Tc. This is understood i f we consider how the 123-Ox structure reacts to pressure. A s discussed earlier (sect. 2) the coordination polyhedra o f the metallic atoms in the 123-Ox structure are the CuO5 pyramids o f Cu2, the YO8 polyhedra between two

100

E. K A L D I S

6,5

12 11" 10" 9 a. 8-

t~

7.

n

54.

~6

, i

6,7

6,8

6,9

7,0

i

i

J

i

i

. . . .

. . . . .

. . . .

. . . .

,

123-0 X

B u c h e r et al. ~ 1991

oe

~o3 "O

6,6

. . . .

B u c h e r et al. 1996

o

dD

•

o

•

•

210'

, i

12 11 10 9 8 7

. . . .

6,5

i

. . . .

6,6

i

. . . .

6,7

i

. . . .

6,8

i

. . . .

6,9

oxygen content x

i

,

7,0

Fig. 52. Giant pressure effect in the pressure dependence of T~ as a function of oxygen nonstoichiometry x in YBa2Cu30~.. After data of Bucher et al. (1990-1992, 1996).

neighboring superconducting planes, the BaO10 polyhedra and the CuO4 squares of Cul. ± of the unit cell volume lies between the two neighboring superconducting planes, 4 but only half of this volume is filled with the YO8 coordination polyhedra. Thus, the compressibility along the c-axis should be appreciably higher than along the a- and b-axis. This is exactly what the experiment finds. At a pressure of e.g., 10 GPa: A V / V = 6 % , Ac/c = 3.0%, Aa/a = 1.4%, and Ab/b = 1.7% (Schilling and Klotz 1992). The question arises what pressure effects on the superconducting properties are expected when the oxygen nonstoichiometry changes. With the removal of oxygen from the chains the apical bond increases so that the coupling between chains and planes decreases. This leads to an increase of the c-axis and increased compressibility in the c-direction, as well as a decrease of To. Application of pressure, therefore, leads to a decrease of the apical bond and increase of Tc. This has been shown to apply, e.g., for the double chain compound YBa2Cu408 (Kaldis et al. 1989, Nelmes et al. 1990). Most interesting is that the pressure dependence of Tc, dTc/dR does not change linearly with nonstoichiometry but shows two giant anomalies in the underdoped regime at x ~ 6.7 and x ~ 6.4. The first effect can be seen in fig. 52 increasing dTc/dP by a factor of 22 (.I) from 0.5 K/GPa at x ~> 6.9 to 11 K/GPa at x = 6.75. The maximum lies between two different oxygen content regions, one with a small pressure effect, 0.5 K/GPa, corresponding to the oxygen-rich, high-To samples (90K) at x/> 6.9 and one with 4.5-5.0 K/GPa corresponding to the lower To plateau (60 K) with less oxygen (6.50 < x < 6.66). This giant pressure effect appears in the narrow range where the To vs. x curve has the highest slope and the onset of the diamagnetic transition vs. x (fig. 2, x=6.75) shows a large step at this composition, where a phase boundary appears (cf. sect. 5.1.1). Figure 53 shows the earlier measurements by Murata et al. (1989) and Medvedeva et al. (1990) (up to 8 K/GPa). Although the high-pressure effects have been measured by these authors, their oxygen contents were not exactly determined and the homogeneity of their samples has been questioned, so that confusion existed about the exact dependence on

OXYGEN NONSTOICHIOMETRYAND LATTICE EFFECTS 1N Y B a 2 C u 3 0 x 16

i

14-

,

i

,

i

'

J

,

i

=~

,

101

i

123-0,

12-

A

a.

Medvedeva et al, 1990 108-

11.

6 -

i_~ i

6,2

6,3

6;4

625

i

6,6

6:7

Oxygen content x

r

6,8

619

7,0

Fig. 53. First measurements of the dTjdP effect shown in fig. 51: solid triangles, Mm'ata et al. (1989); open triangles, Medvedeva et al. (1990). The stoichiometries of Medvedeva et al. were in error.

nonstoichiometry. Medvedeva et al. (1990) found a wide quasi-parabolic dTc/dP(x) curve and not the sharp maximum found by Bucher et al. (1991). Using high-resolution volumetric oxygen analysis (sect. 3.1.3) and high-quality BAO samples (sect. 3.1.2), Bucher et al. (1990) and Rusiecki et al. (1990) found at first a large increase o f dTc/dP with x (from 0.2 to 4.2 K/GPa). Upon investigation of more samples they soon after (Bucher et al. 1991, 1992) found the m a x i m u m dTJdP change at x = 6.75 (fig. 52) and proposed that an electronic or structural transition is probably the reason for this giant pressure effect. A neutron diffraction study under pressure was performed by Jorgensen et al. (1990c), but unfortunately only for x = 6.60 and x = 6.93, so that the composition range of the giant pressure effect was not probed. Only small changes of the lattice parameters were observed for the compositions investigated. However, it is possible that even if the correct compositions are investigated a structural effect - appearance o f superstructures or a phase transition - cannot be observed at the macroscopic length scale of neutron diffraction (/>500 A). In fact, micro-Raman measurements with their small length scale (sect. 5.5) do show a strong change in the behavior of the apex phonon at x = 6.75. Almasan et al. (1992) confirmed and expanded the experimental results of Rusiecki et al. (1990) and Bucher et al. (1990) using single crystals and also investigating lower oxygen contents. They found, however, as did Murata et al. (1989), that at lower oxygen content (x < 6.4) a further increase of dTJdP takes place up to 7 K/GPa and considered a step-like curve with two plateaus (fig. 54). Using the pressure-induced charge transfer model (PICTM) they proposed an inverted parabolic dependence of T~ on Anh. A problem of this phenomenological model is that it considers only the linear charge transfer along the c-axis, while the results of thermal expansivity (Kraut et al. 1993) and also theoretical considerations (Pickett 1997) show that the in-plane axis behavior is important for the x = 6.75 anomaly (sect. 5.4.1). Bucher et al. (1991, 1992) measured more compositions and were able to show that there is no plateau in the region 6.4 ~<x ~< 6.8, but a very narrow giant maximum (fig. 52). Weber et al. (1992) and Benischke et al. (1992) reproduced these results. Their lower

102

E. KALDIS 0,6,2

, 6,4, , 6,6, , 6,8,

. 7,0, , 7,2,8

123-0

0,6

i

0,6

[3 el

0,4 ¸

0,4

I~

a. "ID

[]

Rousieckiet al

X' oo

"ID 0,2 ¸

0,0

0,2

,

6,2

i

6,4

,

~

6,6

,

i

6,8

•

i

7,0

oxygen content x

,

0,0 7,2

Fig. 54. Pressure dependence of (resistive) dTo/dP as a function of nonstoichiometry (left-hand axis). After Alrnasan et al. (1992). Solid squares, Rusiecki et al. (1990).

value of the dTc/dP maximum (7 compared to the 10.5 k/GPa of Bucher et al. 1991) could be due to the different temperatures at which pressure had been applied. They applied pressure to the sample after it had been cooled to low temperatures in order to avoid oxygen-ordering effects. An interesting result of Benischke et al. (1992) was the differentiation between the "true" pressure effect of their polycrystalline material and the pressure effect due to the weak links. For the large change of dTo/dP with pressure as a function of nonstoichiometry (fig. 52) no explanation could be given. However, the fact that their measurements were made at low temperatures indicate that oxygen-ordering effects of the type found by Veal et al. (1990b) are not responsible for the maximum at x ~ 6.75. To investigate the role of oxygen-ordering effects for the composition x ~ 6.75 the same group (Metzger et al. 1993) performed measurements both on polycrystalline samples and on single crystals prepared with a similar thermal history. Utilizing measurements at low and high temperatures they separated the two main contributions to dTc/dP: the intrinsic pressure effect (7.4 K/GPa) which excludes oxygen ordering via diffusion and is measured at low temperatures, and the oxygen-ordering effect (4.1 K/GPa) by diffusion which can be measured at T > 240 K. Their sum 11.5 K/GPa was higher than the results of Bucher et al. (1991) (10.5 FUGPa) and Medvedeva et al. (1990) (10 K/GPa). They concluded that higher oxygen ordering (inter- and intra-chain) is energetically favored when pressure is applied, the oxygens being pushed together. They also showed that ordering effects due to pressure and temperature appear at T > 240 K, which seems reasonable in view of the smaller volume of the ordered samples. The effect of application of pressure on Tc shows a relaxation to an "equilibrium" value with a characteristic time constant. In view of the complexity of the pressure effect on To, it may be questioned whether the system reaches a true equilibrium state and not a stationary state. In any case, the process

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30:~ ~oooo

!!!i!!!!!!!:!!!

.

,ooo

100

,o

103

. . . . . . . . .

.......

~

~ . . . . . . . . . . . . . . . . . . . . . .

....~ . . . . ~ . ..-

..........................

"Y~..

~

]

"

......

~ ..................... ~ ............. y

123

~:::.!!!!!!!!:!!!!!!!!!!!!!!!!!!!~!:i!!!i:!!!!!!!!!!!!!!!!!!:!i!i!!

..................... 6.5

7

i~ 7.5

.~

............

!!

~ . . . . . . . . . . 8

8.5

9

95

Fig. 55. P - T cross section of the pseudo-binary P - T ~ phase diagram of 123-O~. The stability field of the 123 phase is smaller than shown. Slow decomposition kinetics do not allow the exact determination of the phase botmdary with the 124 phase. After Karpinski et al. (1991).

appears reversible and the original To value is reached again after sufficient annealing time at room temperature (Metzger et al. 1993). In a continuation of this work Fietz et al. (1996) expanded their investigation systematically to both the underdoped, x > 6.4, and the overdoped regime using polycrystalline and single-crystalline samples. It is rather doubtful whether the claimed x = 7 has really been reached, in view of the lack of a high-resolution direct oxygen determination. Reduction of the samples was performed by annealing at various temperatures and partial pressures of oxygen, and the oxygen content was always estimated thermodynamically. To bypass the difficulties caused by the extremely slow oxidation of oxygen-rich single crystals they annealed under 200 bar oxygen pressure. The same was done for the oxygen-rich polycrystalline samples. As this method is now generally applied in order to increase the low oxygen content of single crystals, it is worth noting that, as shown by Karpinski et al. (1991) in the quasi-binary P - T - x phase diagram of YBa2Cu3Ox (fig. 55), this pressure lies outside the stability region of 123-Ox up to temperatures of 1050°C. Thus, depending on the temperature and duration of the annealing as well as the rate of the diffusion kinetics, it is possible that this treatment introduces into 123 various defects, e.g., double chain fragments, which are the structural units of the 124 and 247 phases (Kaldis et al. 1991). To our knowledge, no systematic characterization by EM and highresolution synchrotron radiation methods (e.g., Qadri et al. 1997) of such pressure-loaded crystals has been done up to now.

104

E. KALDIS 10

61aa i l

•

j

i

A

a

• Fietz et al. AD 'fang Veal et aL Jorgensen et al

,

i ~ Pressure change

"

60 I

,-.~'sg I,.-u

i

to o,2 GPa

L

_-

= '~

Pressure changeto

~ O,10Pa Pressure change to k . .

58

• . Pressure change to ~"~-ambient pressure

0.3 GPa

~'5'o

Storage

~;o

i~o

%

1; [h]

4

T at ambient pressure

~7

6

200 250

t i m e a t T --- 2 9 8 K

[h]

Fig. 56a. T~ thermal relaxation at 298 K of a singlecrystal 123-O6.58. For every pressure, To is measured at least 8 times after different storage periods at 298 K. After Fietz et al. (1996).

" ...,..... ""...,... I

6,4

6;

61o

o17

Oxygen content x

Fig. 56b. Dependence of the time constant on the oxygen stoichiometry. The curve follows eq. (2) of Jorgensen et al. (1990b) and Veal et al. (1990b). Open squares: Yang (1992). After Fietz et al. (1996).

Thermal relaxation of the samples under pressure at room temperature (Fietz et al. 1996) for sufficient time gives the total pressure effect consisting of the intrinsic and the oxygen-ordering pressure effect dTc _ (dTc~ \~Ji

(dTc~ +\dPjo

The thermal relaxation to the equilibrium value under a given pressure, T~,ei is shown for a single crystal with x ~ 6.58 in fig. 56a. After each pressure change at least 8 annealings at room temperature were performed and the Tc was measured. The sample remained under pressure through all the thermal cycling. Figure 56b shows the time constant T of the T~ relaxation under pressure at T = 298 K as a function of oxygen stoichiometry and the agreement with the relaxation law (2) of the thermally quenched samples by Jorgensen et al. (1990b) and Veal et al. (1990b) discussed in sect. 3.2.1.5. The trend is clear: r decreases strongly with increasing oxygen content. Figure 57 shows the changes of the pressure-induced To changes (dTc/dP) with oxygen stoichiometry for annealings at 298 and 100K (Fietz et al. 1996) in order that intrinsic pressure effect and relaxation effects due to oxygen ordering can be differentiated. Two important results can be seen: (a) The anomaly at x ~ 6.7 appears at both temperatures (appreciably reduced at 100 K) indicating that it has also a non-diffusional component. From the 11 K/GPa at T = 2 9 8 K , 6K/GPa remain at T = 100K. (b) A giant pressure effect appears at x < 6.5 (30K/GPa) which is probably due to diffusional oxygen ordering and, therefore, does not appear at lower temperatnres.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30 ~ 35

2g

i

' F i e t z el al.' 19~)6 ' ' ' [ ] single cryslals Ts < 100 K t

i,

30

•

single crystals Ts = 298 K J

•

potyerystals Ts = 298 K ]

polycrystals

Ts < 100 K /

dTo/dp

Fietz et al. 1994; .= Benischke et al. 1992; D'yakonov et al. 1996 O polycrystals Ts < 100 K

[K/GPa]

•

25

4.

polyc...ystals

Ts = 298 K-

•

6,4 ,

'

6,5 ,

,

6,6 ,

,

105

6,7 , 6,8, . 6,9, . 7,%0 18

18• 16-

"~ ~

14-

~

Fietz et al. 1996 16 14

m

I1. t ~ 12-

12

10o.

-~

"~

10

~

8

~,

"

864-.

i

2 -.

/I',,

10

t

•

•

,

,

',

' b._l.- I

T , < 1OO K

,

, 6,4

6,5

~

,.' ~

"/

....

I 6,6

,

'.,

\

l

O

• • •

/

=

m

i

4 ,,

2

0 . . , . , , . , . i-"-- '0 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0

i

T,=298K

6,3

6

i

15

.

6,7

Oxygen content x

L,

",

I 6,8

Fig. 58. Pressure dependence of Tc due to oxygen-ordering effects as a function of nonstoichiometry of 123-O,. After data of Fietz et al. (1996).

. 6,9

7,0

Oxygen content x

Fig. 57. To dependence on pressure as a function of nonstoichiometry. Solid symbols, total pressure effect (intrinsic and oxygen ordering) after annealing at room temperature. Open symbols, intrinsic pressure effect after annealing at 100K. Data from Fietz et al. (1994, 1996), Benischke et al. (1992) and D'yakonov et al. (1996). After Fietz et aI. (1996).

This justifies the first results of Murata et al. (1989) which have shown this effect (fig. 53) as well as the results of Ahnasan et al. (1992) at x <6.4 (fig. 54). The above measurements of dTc/dP were performed at the various To temperatures and then brought to RT so that an additional pressure effect must be considered due to the fixed volume of the He used as pressure medium. Figure 58 shows the effective pressure effect due to oxygen ordering. For x > 6.93 (overdoped phase) no oxygen ordering takes place. With almost 5 K/GPa, oxygen ordering seems to be involved both in the x = 6.75 effect and in the vicinity of the T--* O phase transition, where, with the high number of vacancies in the chains, the effect becomes very large. The particular properties of the x = 6.75 composition will be discussed in sect. 5.4.2. The giant pressure effect reported by Fietz et al. (1996) at x ~ 6 . 4 is 30K/GPa, a factor of 2 larger than that found by Murata et al. (1989) at the same composition, and by far the largest pressure effect on T~ ever found. It illustrates dramatically the importance of oxygen ordering under pressure for Tc. It is peculiar that this effect is scarcely known. It is reversible, disappearing via a relaxation process (like that of fig. 56a) when the pressure

106

E. KALDIS

is reduced to ambient. Ordering by diffusion in the T - O phase transition (cf. sect. 3.3.1) indicates rather a second-order transition than a martensitic transformation. 5.4.1. Uniaxial pressure investigations dTe/dPi as a function of nonstoichiometry Further proof for the complexity of structural effects in 123-Ox, as well as an alternative explanation for the giant effect at x = 6.75, came from uniaxial pressure investigations of dTc/dP via thermal expansion experiments (Meingast et al. 1991, Kraut et al. 1993) or direct uniaxial pressure in magnetization experiments (Welp et al. 1992). Particularly Kraut et al. performed their investigations at four different oxygen contents. For a second order phase transition like that of the normal ---+superconducting state the pressure dependence of the transition temperature is related to the anomalies of the thermal expansion and the specific heat via the Ehrenfest relation: d~

Aa~Vm

de i

ACp/T c '

with Aai = a s - an the difference of the thermal expansivities in the normal and the superconducting states at To, ACp/Tc the difference of the normal- and superconductingstate specific heats at To, and Vm the molar volume. Thus, by measuring A ai for the three crystallographic axes i = a, b, c and knowing the specific heat and molar volume, the uniaxial pressure dependencies of To can be calculated. One can reconstruct also the hydrostatic pressure effect by adding the three components dTc/dP i along the three crystallographic axes. Meingast et al. (1991), and Kraut et al. (1993) measured the thermal expansivities of single crystals in the three crystallographic directions, using ultrahighresolution capacitance dilatometry. Welp et al. (1992) measured directly the effect of the uniaxial stresses on a.c. susceptibility and confirmed the results of Meingast et al. (1991) (calculated from the thermal expansions) for one oxygen-rich single crystal. The main results of these investigations are: (a) The uniaxial pressure along the c-axis follows the pressure-induced charge transfer model (PICTM) - pressure increases the carrier concentration in a similar manner as oxygen, dTc/dPc is, therefore, strongly dependent on oxygen content and decreases with increasing concentration of holes, nh. Near the optimal doping it reaches a value of zero, then in the overdoped region negative values, because further increase of carriers decreases T~. This agreement with the charge-transfer model between chains and planes due to doping is expected, because as mentioned earlier, the largest changes in bond length appear, both for oxygen doping and under pressure, for bonds parallel to the c-axis. (b) PICTM cannot explain why this is not the case for the calculated dTc/dPa and dTc/dPb. They show opposite sign and have large values at high oxygen content (fig. 60b). By canceling each other they induce the low values of the hydrostatic pressure effect in the overdoped regime (cf. sect. 5.4 and fig. 52). This anisotropy of the a-b plane is clearly apparent from the data of fig. 59 for single crystals. The pressure effect is nearly independent of oxygen doping for the b-axis, in contrast to the c-axis, with the

OXYGEN

a)

NONSTOICHIOMETRY

uniaxial

IN YBa2Cu30 ~

107

pressure

4

b

N

AND LATTICE EFFECTS

o

axis

6~ C axis

"D \ u

~-2 -4-

, 6.5

O_

,

I 6.7

1oI

I 6.8

I

8

C.O

"O \ o I-"o

I 6.6

'

-±

,

a axis I , 6.9

b) hydrostatic pressure

2 0 -2

, 6.5

i 6.6

,

I 6.7

, X

7

I 6.8

,

I 6.9

,

Fig. 59. (a) Uniaxial pressure changes of Tc as a ftmction of stoichiometry in 123-0~, calculated from thermal expansivities of single crystals. We note the strong linear x dependence of dTo/dPc along the c-axis, and the nearly x-independent dTjdP b. dTc/dPc~ and dTJdPb show strong anomalies at x ~ 6.75. (b) Comparison of the directly measured hydrostatic pressure (solid circles, Kraut et al. 1993; open circles, Benischke et al. 1992) and the sum of the calculated uniaxial pressures (solid line, Meingast et al. 1991). After Krant et al. (1993).

exception o f x = 6.75. This striking new feature sheds new light on the discussion o f the above section (sect. 5.4) and indicates that the PICTM is not sufficient to explain these effects, because not only the effects parallel to c-axis are important for Tc, but also additionalpressure efJects. In particular, it shows that dTo/dPa and dTc/dPb are strongly coupled to the orthorhombic distortion at low temperatures and constant oxygen content. (c) Even more interesting is the finding that both dTo/dP, and dTc/dPl~ show large positive values at the anomaly o f x ~ 6.75. We recall that the actually measured data are A a i , i.e. thermal expansivities at the superconductive transition temperature T0.

The expansivity anomalies at To along the a-axis, change their sign Jbr samples with x ~ 6.73 as shown in fig. 60. Thus, at this composition the a-axis expands at To, like the b-axis, whereas in all other measured compositions it contracts. We recall that the a-axis is sensitive to the formation o f superstructures, and as we have seen (sect. 5.1-5.2) this anomaly lies near the boundary o f the 3a0-a0 transition. This would be consistent with the idea that structural effects trigger the large and abrupt oxygen-dependent change o f dTo/dP, (from - 2 to +2 K/GPa at x ~ 6.75, in fig. 59a) (Bucher et al. 1991, 1992, 1996, Rusiecki et al. 1990). A more specific explanation by Kraut et al. (1993) is discussed in the next section.

108

E. KALDIS a) 20

b)

c a x e s (twinned)

a,b a x e s

(untwinned)

45

AP'73-1 x=6.94

dd~p~

~

=

x=6.93 b axis

30

L

AE73-3 x=6.gl

:1.5

10 0

0•1•1 L,

u

o 80

I

80

g0

100

1.2

miAE73-4 x=6.81

AE146

i

=~

x=6.73

i

90

100 a

.. ~ " 1

axis

A444~A~A-- ~4~44~uh4 AA~J. 4 16

I

0

AE73-6

I

50

I

I

70

,

go

110

8

AE187

~,o

x=6.59

nuouamnu~

ill I I I I

5 AEI ~ 2~ ~

°oo~_. , ~ _ ~ ,

~

ua

II

kA&AAAAAA4

a

n

50

I

60

T(K)

e

n 70

IN II

&AA~" "AA AAAA

x=6.63

0

b axis

i

l

50

60

sxis

i

70

T(K)

Fig. 60. Anomaliesof the thermal expansivitiesat Tc (a) along the c-axis and (b) along the b- and a-axes of single crystals of 123-O., after subtractingthe background.Note the sign changes of the a-axis at x ~ 6.75 and of the c-axis near optimal doping. The scales of the vertical axes are not the same and some data are shifted vertically for clarity. After Kraut et al. (1993).

5.4.2. The structural, optical and magnetic anomalies at x ~ 6. 75 After all these investigations the question arises again what is the nature of the effect at x = 6 . 7 5 (Bucher et al. 1991). Fietz et al. (1996) exclude a diffusional process because oxygen motion effects have time constants which increase strongly at T ~<240 K. This is supported by findings in the literature (sect. 3.2.1.5). Thus, Veal et al. (1990b) found for an x ~ 6.45 sample an increase of the time constant from 0.25 to 229 days already by decreasing the annealing temperature from 25°C to 0°C, and Lavrov (1992) with conductivity found an abrupt onset of oxygen ordering above 250 K. However, latent structural and/or electronic phase transitions or cross-overs can induce structural and electronic anomalies even at very low temperatures. This explanation was proposed by

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN YBa2Cu3Q

109

Bucher et al. (1990-1992, 1996) and (Rusiecki et al. 1990); ideas in the same vein were presented by Metzger et al. (1993) and Fietz et al. (1996). They consider atomic displacements or tilting of structural elements with a low activation energy as the possible reason for this effect. Pressure-dependent reorganization of the charge distribution due to ordering of chain fragments was proposed by Kraut et al. (1993). We summarize here the anomalies of the structural and physical properties accompanying the giant dTc/dP change at x=6.75: (a) Lattice parameters. From figs. 25a,b (sect. 3.2.2.1) we see that the deviation from linearity takes place in both cases near x = 6.75, leading to a relative decrease of the b-parameter and a relative increase of the a-parameter. These changes culminate in the increase of tetragonality at x = 6.95, where the displacive phase transformation takes place (sect. 6) and the b-axis decreases. The corresponding decrease of the orthorhombicity indicates that, possibly, already at x ~ 6.75 the negative interactions start, leading to the decrease of To in the overdoped phase (see also point c). The suggestion that the results of fig. 25 could be due to an increase of the 05 occupancy (Krfiger et al. 1997) does not seem to be supported by EXAFS. Also, von Zimmermann et al. (1999) did not find 05 occupation. (b) Magnetization measurements/Superstructures. It is well known that the Tc vs. x curve shows a strong change of slope from the so-called 60 K to the 90 K plateau at x = 6.75. The abrupt change of the onset temperature (fig. 2) illustrates this even more clearly. As discussed in sect. 5.1.1 this is the locus of a miscibility gap or of several phase boundaries (cf. fig. 109), supporting the above-mentioned idea of a specific feature of the structural phase diagram. A hard X-ray phase diagram (fig. 44b) shows, in this oxygen content range, the 3a0-a0 phase boundary. These superstructures have different a-axis lengths. (c) Raman measurements. An important change takes place at x ~ 6.75 in the apex frequency, sect. 5.5.1 (fig. 61). In the range 7. 0 ~ x ~ 6. 75 the apex phonon frequency becomes independent of the oxygen content, in contrast to the underdoped region. Liarokapis (1997, 2000) measured the Fano asymmetry of the in-phase Blg phonon as a function of the oxygen doping. This asymmetry results from a coherent interference of light scattered by the continuous free carrier spectra and the discrete phonon spectra. The coupling to the carriers (fig. 35c) peaks at x ~ 6.7. Both results support the above-mentioned beginning of overdoping at this composition. Liarokapis et al. (2000) measured also the anharmonicity of the oxygen phonons in the whole range 6.03 <x < 7.0, and found that particularly for apex the maximum appears at x ~ 6.7. This will, possibly, activate the discussion about polaron formation. (d) Pressure-dependent charge redistribution. Pressure-induced ordering in the basal plane cannot be concluded from the results of Kraut et al. (1993) since they have not been obtained by application of pressure. They adopt an alternative explanation for the anomalies of their thermal expansivity measurements and their calculated uniaxial pressure changes of dTc/dP along the three crystallographic axes. Short chain segments have energy levels above the Fermi level and, therefore, they cannot contribute to the charge transfer (Burdett 1992). For chains with a critical size

110

E. KALDIS

(x ~ 6.73), the energy bands may shift strongly with pressure, leading to a large pressure-induced charge redistribution, which takes place mainly along the a- and b-axes, being thus different to the commonly known charge transfer along the c-direction. The above-mentioned presence of the superstructure phase boundaries may enhance this redistribution. The pressure-dependent charge redistribution is also supported from theoretical calculations of the effect of uniaxial pressure on the band structure of stoichiometric YBa2Cu307 (Pickett 1997). The results do not show any charge transfer from the chains to the planes under pressure, which is in agreement with the very small pressure effect on Tc for this stoichiometry. There is, however, a charge redistribution inside the superconducting planes due to relative changes in the potentials of the Cu2, 0 2 and 03 ions. It is generally accepted that at low doping levels the holes go to the Cu2 ions and their four O NN (Zhang-Rice singlets). With increasing doping there is a shift of positive charge towards the oxygens, increasing T¢. This trend is consistent with the increasing charge on 02, 03 with uniaxial pressure along the b-axis. In conclusion, the x ~ 6.75 composition is an important point of the phase diagram where structural, magnetic and optical properties show appreciable changes. 5.5. Raman scattering as a function o f nonstoichiometry - I 5.5.1. The apex Raman phonon and phase separation Very interesting in this context are the results of micro-Raman scattering on individual single crystallites of equilibrium samples (Liarokapis 1997, Kaldis et al. 1997b) as a function of nonstoichiometry. Some aspects have already been discussed briefly in sect. 3.3.1. A major advantage of Raman scattering is its great sensitivity to the appearance of new phases as it takes only a few atoms to form a new phonon. The reason is that the frequency of the Raman mode of the vibrations of mainly one type of atoms is much stronger influenced by interactions of these atoms with their nearest neighbors than with their next-nearest neighbors (e.g., Iliev et al. 1993). Thus, an effect appearing in approximately 20 atoms (micro domains < 40 A) can still be detected. This should be compared with the 500-1000 A length scale of the X-ray diffraction methods. Of course this ideal resolution of the Raman scattering may be decreased from case to case due to experimental limitations like, e.g., the laser beam cross-section. Several results of Raman investigations on equilibrium samples have been reported recently (Poulakis et al. 1996, Liarokapis 1997, 1999, Liarokapis et al. 2000, Palles et al. 2000a). The measurements were performed on individual microcrystallites (platelets) using a T64000 triple Jobin-Yvon spectrometer with a microscope, in the scattering configurations yy(zz)~ and y(xx)~. In these directions the z-axis coincides with the crystallographic c-axis perpendicular to the platelets. Looking at individual platelets with polarized light it is possible to select properly oriented crystallites for micro-Raman measurements. For 29 compositions in the range 6.020 ~<x~< 6.984 several (up to 10) crystallites have been studied from each composition.

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN YBazCu30~.

111

An advantage of the micro-Raman method is that probing very small crystallites (>5 ~tm) or several parts of larger crystallites can give a good statistic of the various phases appearing in a certain composition range. Owing to the very cautious cooling of the samples used to reach equilibrium and their very good reproducibility, variations of the Raman signal could be attributed largely to phase changes. An appreciable statistic, the investigation of a large composition range, and the high-resolution determination of the oxygen content allowed to differentiate changes of the signal due to reproducibility and due to phase changes. Thus, it is more instructive not to take the average signal of all crystallites having the same composition, but to plot the value of each individual measured. This shows possibly existing centers of weight inside the normal scattering of the data, indicating the existence of other phases. Thus, in addition to being very useful for the characterization of the phase purity of the samples, inspection of the individual data allows scouting for possible phase separation or phase transitions occurring at certain compositions. We will illustrate this point later, especially with figs. 62 and 64. In the following figures all measured data are shown. This enables the reader to have a realistic impression of the degree of homogeneity that even good samples have. The samples used in these investigations were CAR equilibrium samples (sect. 3.1.2) which show extremely small scattering in their lattice parameters (sect. 3.2.2.3). An important message emanating from these investigations is that the Raman phonons of equilibrium samples show appreciable deviations from the linear dependence on oxygen doping found in the past (for reviews, see Thomsen 1991 and Thomsen and Cardona 1994). Figure 61 shows the Raman frequency shifts of the apex phonon as a function of the nonstoichiometry (Poulakis et al. 1996, Liarokapis 1997, Liarokapis et al. 2000, Palles et al. 2000a). Contrary to earlier investigations (MacFarlane et al. 1988, Thomsen and Cardona 1994) which found a linear dependence using a rather small number of compositions, a nearly linear increase with oxygen in the insulating range exists only up to the RT N6el transition ( 6 . 0 < x < 6 . 2 , cf. sect. 3.3.1) and then in the superconducting range only for 6.4 <x < 6.75. The hardening of the apex phonon is due to the decrease of the apical bond with oxygen doping (cf. figs. 14, 66). Above the phase boundaries at x = 6.75 the apex frequency becomes independent of the oxygen doping. In this range the charge transfer nears saturation and the frequency remains constant, although the Cu2-O1 bond length is still decreasing (fig. 14). The structural changes starting at x > 6.75 were discussed in sect. 5.4.2. A relative decrease of the orthorhombicity starts at this doping level which is actually leading to a saturation in the optimally doped region and to an absolute decrease in the overdoped range, mimicking Tc (fig. 25c). Also, the width of the out-of-phase-phonon Big shows a decrease of the coupling with the superconducting carriers starting at x = 6.7 (fig. 35c), indicating the onset of negative interactions. Figure 61 shows that the apex phonon frequency clearly reflects the transformations and changes at x ~ 6.22, 6.28, 6.40 and 6.75. The first three are correlated with the onset of superconductivity, as discussed in sect. 3.3.1. Figure 62 shows the width of the apex phonon vs oxygen doping after Poulakis et al. (1996), and Liarokapis et al. (2000). The lines are a guide to the eye and are joining the extreme values. The statistical average for each composition is shown in fig. 35b.

112

E. KALDIS 506

5,9 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7 0 7,1 .... , ' " , u ' " ' ~ =. . . . . . . . I , , , ' , ' " ' u , • , i . . . . . . . . . . 506

o

•

~

m

504 -

-

AF

- 504

502 -

E

~

•

o s00-

II=Z'

II

~.

0-

498O

~. 496

~

!

)

"502

•

- 500

1

•

-'

I-"

-

---

496

o

492-

4kA

la/

490-

~-

C •

"o o¢J

(11

-

480

• T=300K Micro-Raman

476

....

- 486 - 484

p-

- 482

Q_ u)

•

478

-~-- 488

X < II n

8

O 486~ " 484 ;

490

b

E 488-

~ '482

•

" ~ - 494

494 -

~

498

Fig. 61. Dependence of the apex phonon (Cu2O1, Ag) fi'equency on oxygen stoichiometry. Five regions appear to exist (cf. fig. 35a); the slope changes take place at the characteristic oxygen contents x=6.22, 6.30, and 6.75. After data of Poulakis et al. (1996), Palles et al. (2000a) and Liarokapis et al. (2000).

- 480 - 478

o~ o

I''''f''''l'''

476

-

I .......

I''''1''''1'

5,9 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 7,1

oxygen content x

"7

45 1[1 . . .AF-in..~ulator . i . . . . i . . . . i ,,,J . m . . . i . . . .underdoped i . . . . i . . . . i . . . . i . . . . . . .~. . . o 40

c~

> )

,."! "~

35

' • ~ilm•

iN

~'i

!1

:m • :n qmm •

:

"i

: 3o

~:"

~} 25

20

'

/mmm ~=

6,0

6,1

6,2

.

=

6,3

•

I

mm m"l

.

:

.' "-m ~ :

m,:

unl

" m, • F-.mm U

•

• fil I-..m,,

m"

"'

i-J=~ ............... Liarokapis et am:j

. . . . , . . . . , . . . . ,.

5,9

o .~ ~'i... co = ~:;m l

" .• k

6,4

Oxygen

6,5

6,6

content

6,7

.

.

.

6,8

x

.

.

.

.

6,9

.

.

.

.

7,0

.

7,1

Fig. 62. Apex phonon width as a function of nonstoichiometry. Note the two maxima at x-6.35 and 6.56 and the minima at x=6.0-6.15, 6.43, and 6.95 6.99. After Poulakis et al. (1996), Palles et al. (2000a) and Liarokapis et al. (2000).

The a n o m a l o u s width o f the apex p h o n o n led previous investigators (Iliev et al. 1993) to deconvolute, with three Lorentzians, the spectra o f one u n d e r d o p e d single crystal (x ~ 6.4). They c o n c l u d e d that phase separation at mesoscopic scale was present and attributed the Lorentzians with peak intensities at 500, 488 and 474 c m -1 and constant width to the coexistence o f the ortho-I, ortho-II and tetragonal phases. Poulakis et al. (1996), Liarokapis (1997, 2000), Palles et al. (2000a) and Liarokapis et al. (2000) investigated a very large n u m b e r o f crystallites o f m a n y C A R e q u i l i b r i u m samples in the

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa~ Cu 30x

113

0.5

ooJ

~

m

0.5 "~ 0.0 "J-~F +±

~m---~l~

o.0 ~

~

~ 1.o

o.51~

~

475cm -1

0.0 --J J_ ~ ' ~ ' ~ [

6.0

'

I

6.2

.m-'~Ull

~k'

I

6.4

'

I

I

6.6

6.8

oxygencontent

'

I

7.0

Fig. 63. Deconvolution of the apex phonon width in four Lorentzians with 502, 493.5, 486 and 475 cm i (see text). After Palles et al. (2000a) and Liarokapis et al. (2000).

whole nonstoichiometric region (6.0 < x <7.0). After trying several fits they concluded that the experimental curve could be very well fitted with four Lorentzians with constant width independent of doping. The deconvolution gave for the four Lorentzians the frequencies 502, 493,486 and 475 cm -1. The relative intensities of these four apex phases are shown in fig. 63. At each oxygen content the sum of the four intensities is 100%. Figures 35a and 64 show the above frequencies (horizontal dotted-dashed lines). From these figures the composition range of these phases can be approximately derived. It can be seen that: - The 502 cm -I mode dominates the oxygen-rich superconducting regime (optimally doped, overdoped) where the other modes are practically zero. It can be, therefore, assigned to the stoichiometric phase with ortho-I structure (a0). - The 493 cm -1 mode has its maximum intensity at x = 6.47, which is consistent with figs. 35a and 64. It extends its maximum in a large range which is consistent with the ortho-II phase (2a0 superstructure) in the phase diagram of fig. 44b, and with the lattice parameters (figs. 25a,b). - There is no indication that the 486 cm 1 mode corresponds to the ortho-II phase as proposed by Iliev et al. (1993). As shown in fig. 35a this phase comes in resonance by red excitation in the range 6.3 < x <6.4. In some samples traces of it appear up to x = 6.56 (figs. 35a, 64), and it seems to show magnetic response (Liarokapis et al. 2000). Thus, it is probably not the ideal tetragonal, but rather a coexisting disordered phase with rather wide stability range. Samples treated with a 5.5 T magnetic field

114

E. K A L D I S '',1

....

t ....

i;,,.ii

....

[ ....

i ....

i ....

i ....

~ ....

i ....

i,,

..... 505-. .

.

.

.

............

"70 E

500i

', -i . . . . ~ . . . . . . . .

476.2 cm-' i ! ~ / i

o>" 4 9 5

~

@ 490 '~

2"

48o

r

475

y "

---

, ,, i

485:

.:-- "

•

:

•

: 1 2 3 - O× ,

|!

Liarokapis

i:

eta,

.... , .... , .... ,....',. . . . . . . , .... , .... ,..'.., .... , .... ,... 5,9 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 7,1

oxygen content x

Fig. 64. Coexistence o f apex phases as seen f r o m the R a m a n shifts. With the exception o f the f r a m e d data, w h i c h were obtained with blue laser excitation (476.2 c m i), for all other data g r e e n laser excitation ( 5 1 4 . 5 c m -I) w a s b e e n used. For details see the text. After data o f Liarokapis et al. (2000).

show a decrease of frequency in the range x < 6.5 where this phase dominates. It could possibly be the phase of the hole-doped clusters (sect. 3.3.2). - The 475 crn-1 could be the ideal tetragonal 06 without oxygen in the chains. It coexists near 6.0 with the 486 cm -I and disappears at the I-M transition (x ~ 6.4). The apex phonon frequency vs oxygen content is a good example to illustrate the additional information that can be acquired about the material by the inspection of individual data as compared to their statistical average, provided that material with high reproducibility is used (like the CAR samples, sect. 3.2.2.3). Figure 64 shows the apex phonon-frequency data of the green (514.5 cm -1) and blue (476.2 cm -I) excitations (cf. fig. 35a). The latter are fi'amed for recognition. Even if we discard the green excitation data at the same compositions with the blue excitation, the splitting of the individual data of most samples shows the coexistence of two phases, whereas fig. 35b gives no such indication at all. In conclusion, Raman investigations show the coexistence ofseveralphases in 123-Ox, with fractions changing as a function of the nonstoichiometry. As these investigations probe the local level of the atoms, this effect is not due to a macroscopic phase separation obeying the T - x phase diagram and discussed in sects. 5.1 and 5.2. It is a microscopic effect showing rather a kind of physical phase separation in phases only partly represented in the macroscopic phase diagram. Physical phase separation has been discussed since many years, particularly for the La cuprates (Mfiller and Benedek 1992, Sigmund and Mfiller 1994, Mihailovic et al. 1995, Kaldis et al. 1997a, Bianconi and Saini 1997, 2000). 5.5.2. The steps in the in-phase phonon frequency Figure 65 shows the individual crystallite measurements of the 02, 03 in-phase vibration mode parallel to the e-axis (Ag phonon) as a function of the oxygen content. In this section we will discuss only the effects appearing in the underdoped phase 6.03 ~<x ~<6.89. The results of the overdoped phase will be discussed in sect. 7.4, and the T - x phase diagram in sect. 9. The in-phase phonon shows a large dependence on the oxygen content of the chains (doping due to charge transfer), its frequency decreasing strongly with

OXYGEN NONSTOICHIOMETRYAND LATTICE EFFECTS IN YBa2Cu30~

115

6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 . . . . . . '11 . . . . i ,,i ,~ .,~ 458

•,-.., 466- •

,

1i

'\

°' i

454" 452- ~

k.,,

~'i

450448 I~

0

I"" .~

452 450

.

. °

'-2t

442 "~

436 -

-

x

-- ~" -

' o

~

~

~'~m

436 -

0

c~

>-

~ 448 --

4400

Ei~ ~

-

446444-

O ~'

° 3 I ,,la °

IIIli •

I II'

f [-440

~1-438 L4a6

434 :i --- ~ f 434 . Raman, $ 432 - r = 300 K E 1-432 "Poulakis et a1.1996 430 430 Palles et al. 2000 428 I 428 , . , i . . . . i . . . . i . . . . i ,. i . . . . i . . . . i, . . . . . . . . 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0

f

-

oxygen

content

x

Fig. 65.02,03 in-phase vibration (Ag phonon) vs. oxygen doping, measured with microRaman on individual monocrystallites. A strong softening of the phonon appears in the overdoped range where neutron diffraction (Conder et al. 1994a) has shown a discontinuity in the Cu2-O2,O3 bond (dimpling) (c£ fig. 66a). The trends are visible, showing phase boundaries indicated by vertical and horizontal lines (see text). After data of Poulakis et al. (1996) and Palles et al. (1996, 2000a).

increasing x (Poulakis et al. 1996), as the dimpling increases. A shift of ~6% (456--~ 4 3 0 c m -I) takes place in the range 6.03 ~<x~< 6.984 with increasing oxygen doping as the dimpling increases (fig. 66a, sect. 6.2). A sharp softening of this phonon takes place in the overdoped region (sect. 7.4). In contrast, an increase in frequency o f ~5% (478 ~ 503 cm 1) takes place for the apex phonon in the same range (6.0~<x~< 7.0) (fig. 61) as the apex bond decreases (fig. 66b). As mentioned above (sect. 5.5.1), for x ~> 6.7 the frequency of the apex phonon remains constant. The antiferromagnetic transition at x = 6.22 is clearly seen with a discontinuity, and from this doping level the frequency starts to decrease with increasing oxygen content (fig. 65). This is consistent with the conclusion drawn in sect. 3.3.1 that doping of the planes starts at x = 6.22. Similar to the in-phase phonon width (fig. 35d), the frequency also shows a strong splitting of the data o f each sample in the 6 . 2 < x < 6 . 4 range, indicating the coexistence of phases. As can be seen from fig. 65 (cf. figs. 110, 111), the Raman shifts of the 02, 03 in-phase mode (characteristic for the dimpling of the planes, sect. 6.5, figs. 76, 77) soften stepwise for dopings x > 6.4, forming, with increasing oxygen content, a descending staircase with step height 2 - 4 c m -1 (Kaldis et al. 1997b). As the masses o f the atoms involved do not change, a straightforward explanation would be a stepwise increase of the distance between the oxygens of the planes O2,O3 and Cu2. The increasing trend o f Cu2-O2,O3 distance with oxygen content has been shown by neutron diffraction (Conder et al. 1994a, Hewat et al. 1991) (fig. 66b) and confirmed by EXAFS measurements (R6hler et al. 1997a,b, Kaldis et al. 1997b) (figs. 76, 77). But local probe measurements (e.g., EXAFS)

116

E. KALDIS

at compositions sufficiently near to prove possibly existing stepwise changes of dimpling are not yet existing. It is very interesting that the compositions at which the softening steps appear in fig. 65 correspond approximately to the ideal oxygen contents of a nao superstructure series (Kaldis et al. 1997b), the first members of which were found in the past by ED and HREM. Interactions of the plane oxygens with the chain occupancy is a very complex subject and presently not easy to understand. Anyway, the experimental fact remains, so that an interaction between the stepwise softening of the dimpling in the planes and the ordering of the chains should be considered. A staircase of "dimpling-chain superstructures" which could become visible due to the small length scale of the Raman scattering and the high resolution of the oxygen content, would support and extend the picture of the various superstructures of the underdoped 123-Ox, which has been discussed from the early days of riTe superconductivity (sect. 5.1.1). The staircase of fig. 65 suggests the superstructure series nj 2 + nj 1 = n j (2a0, 3a0, 5a0, 8a0, . . . , nao), which in the picture of classical thermodynamics should be separated by miscibility gaps (two-phase regions, e.g., 2a0 + 3a0, 5a0 + 8a0) or phase transitions (3a0 -+ 5a0). We note that such superstructures, if they existed, would probably have, above 3a0, different chain sequences than those experimentally found by von Zimmermann et al. (1999) (sect. 5.1.1). However, no direct structural evidence exists up to now for the nao series with n > 3. In principle, it is possible that only the corresponding phononic and electronic states exist, leading to the steps of the in-phase phonon at given doping levels, without measurable change of structure. It is also possible that these steps are associated with the organization of stripes. The in-phase phonon frequency steps appearing at the compositions corresponding to the superstructures are shown in figs. 110 and 111 with an extended x scale (see sect. 9). The data are divided by horizontal and vertical lines defining the oxygen content of the in-phase steps. Thus, we see that, e.g., the data in the range 6.60-6.75 (fig. 110) have a scattering of ~3 cm -1 (442-445) which is practically independent of the oxygen content and the origin of the batch. There are also data in the range 6.42-6.50 with a scattering of 2 c m 1 (446-448) independent o f x (in the range of 1 cm-1). The actual steps appear at 6.50 and 6.56, rather clearly indicating the two-phase field (miscibility gap). The assignment to the superstructures offers itself, as these fields coincide fairly well with the compositions corresponding to the above superstructure series. These are indicated by the asterisks in the lower part of the diagram. Thus, the phase fields we just considered would correspond to the 2a0 and 3a0 superstructures and a miscibility gap between them. These two superstructures have been detected originally with ED and were made visible with the nanoscale resolution of HREM. In fact, these and other chain-superstructures have been detected by HREM as defects in samples of various chemical and thermal histories 2. Later, the above two first members of this series could be studied in single crystals of the corresponding oxygen contents with XRD and ND (sect. 5.1.1). Figures 110 and 111 imply that, possibly, the high-quality equilibrium samples (sect. 3.1.2) used in this investigation, combined with the high-resolution determination 2 We recall, however, that HREM cannot directly see oxygen and that the superstructure is deduced from changes of the heavy metal patterns which have strong contrast (e.g., Van Tendelooand Amelinckx 1990).

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN YB%Cu30x

117

of the oxygen content (sect. 3.1.3) enabled the Raman investigations to probe such "phase relationships". As we already discussed (sect. 5.1.1), high-resolution diffraction experiments with synchrotron radiation have recently shown (in addition to 2a0 and 3a0) the existence not of tile 5a0 and 8a0 superstructures with every 5th or every 8th chain empty (which would indicate very long range interactions), but of the ortho-V (2a0 + 3a0) and ortho-VIII [(2ao+3ao)+3ao] superstructures with chain sequences (10110) and (10110110) in bulk crystals (von Zimmermann et al. 1999). With the exception of 2a0, the other superstructures appear in small 2D domains. It should be stressed that the structural work up to now could resolve only chain ordering. It is the Raman results which indicaw the possible interaction between stepwise dimpling changes and chain ordering, as their oxygen contents seem to coincide. We also note that in view of these complex superstructures and their two-dimensional, possibly non-equilibrium character in bulk crystals (von Zimmermaun et al. 1999), it is to be expected that even for small differences of the chemical and thermal history of the samples, different "thermodynamic" relationships may be found. Even small activation energies may pin the system, leading to stationary states which are not necessarily in thermodynamic equilibrium. Differences are expected between microcrystallites and bulk crystals which are much more difficult to equilibrate. With both total history of the samples and length scale of the detection method different, also the compositions which appear in the phase diagram may be different. This is already demonstrated with the differences in the phase sequence appearing in figs. 44b and 65, as well as with the phase diagrams of figs. 110 and 111. An important question which cannot be answered at present is the influence of these "dimpling-chain superstructures" (Kaldis et al. 2000a) on superconductivity. One can speculate that possibly the characteristic dimensions of these superstructures are important for increasing the superconductive coupling. Also, it is possible that the lamellar nanoscale character of the full-empty sequences of chains in these superstructures may lead to the formation of stripes, which have been shown to exist also in 123-Ox (cf. sect. 7.5). Dimpling-chain interactions may possibly be understood better using the ideas of Andersen et al. (1995). In conclusion, whereas in the in-phase 02,03 phonon frequency vs x curve steps can be resolved corresponding to compositions of a chain-superstructure series, no such effect can be seen in the apex phonon. If a staircase of phases existed, as indicated by the in-phase phonon steps, it would divide the superconducting range of 123-Ox in alternating zones of phase-separated material. 6. O v e r d o p e d phase - I: displacive martensitic transformation

6.1. Overdoped phase investigated with equilibrium samples and high resolution volumetric analysis As mentioned already, the overdoped phase has been comparatively less investigated in the past, from the point of view of the materials research. In view of the fact

118

E. KALDIS

that in 123-Ox the overdoped range covers only a very narrow stoichiometric range (AXoverd~ 6.984-6.950=0.034), the difficulty to adjust the oxygen content by (the so popular) quenching, and the uncertainty about the exact oxygen content due to the use of rather crude methods of oxygen determination (Ax =0.1-0.01) have appreciably limited the number of overdoped samples investigated. In most cases Tc is used as a relative measure of x. Using equilibrium conditions for the slow reduction of oxygen in 123-Ox samples with yttrium metal of exact known weight, the oxygen content can be adjusted to better than within AXcom= 0.01 (sect. 3.1.2.1). The use of volumetric analysis (sect. 3.1.3) then allows the determination of oxygen with the extremely high accuracy of Ax = 4-0.001. in this way even in the very narrow overstoichiometric phase a series of samples can be synthesized without particular difficulty. This has allowed a systematic investigation of these samples with structural (average and local structure with XRD, ND, EXAFS), magnetic and optical (Raman) measurements. 6.2. Dimpling (Cu2-02,03) transition at the onset of the overdoped regime." neutron diffraction In a previous section we have discussed the c-axis anomaly found first for BAO equilibrium samples by X-ray diffraction at the onset of the overdoped regime x = 6.92 (sect. 3.2.2.4, fig. 29; Rusiecki et al. 1990). This effect was reproduced by Nakazawa et al. (1994). Possibly, Claus et al. (1992b) may have seen it also (sect. 7.1). In order to additionally verify this effect, and particularly to find the changes of bond lengths corresponding to this anomaly, an investigation of a series of 12 typical BAO samples in the range 6.62 ~<x~< 6.98 was performed by high-resolution neutron diffraction at ILL, Grenoble (Conder et al. 1994a, Kaldis 1997). Figure 66 shows the results of these measurements (Hewat et al. 1991). The corresponding data are given in table 8 (sect. 3.2.2.4). The minimum of the c-axis (fig. 29) could be verified also by neutron diffraction (fig. 29f). It was also found to be the result of an abrupt discontinuity appearing at the onset of the overdoped regime in the Cu2-O2,O3 separation along the c-axis (dimpling), see fig. 66a. This effect is also accompanied by a small anomaly of the apical bond (fig. 66b). We note that the details of the anomaly shown in fig. 66a (NPD) are characteristic for the BAO samples. The corresponding anomaly of the dimpling of the CAR samples (EXAFS, fig. 76) is slightly different. Figures 7 (sect. 2) and 70 (sect. 6.4) show the dimpling of the superconducting planes. The average of the two 0 2 and 03 positions, which increases with increasing oxygen content, changes slope abruptly at x ~ 6.95 (fig. 66a). This abrupt change of dimpling indicates the existence of a structural phase transition at the onset of the overdoped regime. As already mentioned (sect. 5.5.2) and discussed below, this transition was shown to exist also by subsequent Raman (sect. 6.3) and EXAFS investigations (fig. 76, sect. 6.4). We note that according to fig. 66a the increasing Cu2-O2,O3 separation scales with the increasing Tc, up to the onset of the overdoped phase. The scaling of dimpling with Tc has been found also for Ca-123 (B6ttger et al. 1996), a clear additional correlation between structure and superconductivity.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa 2Cu30x

119

6,60 6,65 6,70 6,75 6,80 6,85 6,90 6,96 7,00 E

i

0,280-

8 0,276~123-O 8,

0,2700,265-

BAOsamples NPD,T=5K A.Hewat et al.

i i

a) ')'~ q- /

~/~ ~

i X T y

0,280 ~_

0,275 0,270

~/,1,

±

~ Z

0,265

C

O 0,260, n

t~

0,255-

0,255

0,250.

0,250

0,245

, 0,245 6,60 6,65 6,70 6,75 6,80 6,85 6,90 6,95 7,00 oxygen content x

6,60 6,65 6,70 6,75 6,80 6,85 690 695 700 2,34 . . . . , . . . . . . . . . , . . . . , . . . . , . . . . ' . . . . ' , . . . . ' 234 2,33 ~

.

~

.

b)

2,33

t,'- 2,32-

2,32

2,31 -

2,31

"O i,@ m

,e-

123-O 2,30- NPD, T=5K - BAO samples 2,29- A.Hewat et al. 1991

"k.

2,30 ~ ~

2,28 1"~

,

2,29

7

2,28

2,27 . . . . , . . . . , . . . . , . . . . , . . . . , . . . . . . . . ,... 2,27 6,60 6,65 6,70 6,75 6,80 6,85 6,90 6,95 7,00

oxygen content x Fig. 66. High-resolution neutron diffraction at T = 5 K of BAO samples (sect. 3,2.2.4). (a) With increasing oxygen content the dimpling increases. The dimpling anomaly corresponds to the c-axis anomaly (fig. 29f) at x=6.93, earlier found using X-rays by Rusiecki et al. (1990), c£ fig. 28a. (b) The apical bond shows only a very small anomaly. After Conder et al. (1994a) and Hewat et al. (1991).

6.3. Raman scattering as a function of nonstoichiometry - II. Softening of the in-phase 02, 03 phonon Raman measurements on underdoped and insulating equilibrium samples were discussed in sect. 5.5 and sect. 3.3.2 respectively. In the overdoped region the largest Raman shift as a function of the nonstoichiometry appears for the 0 2 , 0 3 in-phase vibration mode (Ag phonon). As discussed in sect. 5.5, fig. 65 shows the dependence of the Raman frequency of the 02,03 in-phase Ag phonon on the oxygen content. In the optimally doped and overdoped phase a decrease of the phonon frequency by 8 cm -1 ( ~ 30% of the total change) takes place. Near the onset of the overdoped regime (x ~ 6.95) an abrupt softening (45% of the highest value at x = 6.44) takes place for CAR samples, x ~ 6.95 is very near to the stoichiometry of the appearance in the BAO samples of the minimum in

120

E. KALDIS 46,4.6:5

"7,

6:6.6:7.6:8.6,9

7,040

E e.-

123-O, 35-1 Raman, Poulakis et al. 1996

35

/

D

30 25

.~'~oo K / I~

~15-

o

~"

•

2O

=,/77 K

10

10c --

5

6,4 6,5 6,6 6,7 6,8 6,9 7,0

Oxygen content x

Fig. 67. Temperature dependence of the Raman in-phase (02,03) phonon width vs. oxygen content. Note the abrupt increase of the width at 77 K for 6.9 < x < 7.0, indicating the onset of a structural phase transition. After Poulakis et al. (1996).

the c-axis as found by XRD (Rusiecki et al. 1990; see sect. 3.2.2.4, fig. 29) and NPD, and of the abrupt discontinuity in the Cu2-O2/O3 bond discussed above (sect. 6.1, fig. 66a, Conder et al. 1994a). As there is no correlation of this effect with other phonon modes, it becomes clear that the reason must be the structural transition corresponding to the abrupt change of the dimpling (Cu2-O2,O3 bond). This is also supported by the temperature dependence of the width of the in-phase phonon which at T ~< 160 K shows a widening with an abrupt maximum at 6.9 < x < 7 . 0 (fig. 67). A study of the local structure by EXAFS (sect. 6.4) confirmed the existence of this structural transition and showed some important aspects of the structural mechanism. Contrary to the above behavior of the in-phase Ag phonon, the out-of-phase vibration of the O2,O3 (Big phonon) is not affected both in frequency and in width by the change of the oxygen content (fig. 35c). 6.4. Local structure of the planes." EXAFS in the optimally doped and ouerdoped

regimes The EXAFS investigations of the apical oxygen were discussed in sect. 4.4. Here we discuss the local structure of the planes in conjunction with the displacive martensitic transformation at the onset of the overdoped phase. R6hler has systematically investigated in the last few years the local structure of equilibrium samples of 123-Ox as a function of nonstoichiometry (R6hler et al. 1997a,b, 1998) using yttrium K edge EXAFS. The dependence on the oxygen content was measured in the x=6.801-6.984 and T = 2 0 - 3 0 0 K range, originally with 4 (R6hler et al. 1997a), then with 6 (R6hler et al. 1997b) and finally with 10 stoichiometries (R6hler et al. 1998). The results show appreciable displacements of the Cu2 atoms in the superconducting planes, which induce distortions of the 02,03 oxygen and sensitive changes of dimpling, culminating in the transition between the optimally doped and overdoped phases. For optimal doping the distortions of the C u Q planes become smallest

O X Y G E N NONSTOICHIOMETRY A N D LATTICE EFFECTS IN YBa2Cu30 ~

121

and also temperature independent in the superconducting regime (T < 80 K). Also, as a function of temperature clear local structure anomalies were found both at Tc and at the spin-gap opening temperature ( ~ 160 K). The principles of EXAFS have been discussed in sect. 4.4. For the investigation of the local structure of 123-Ox as a function of oxygen nonstoichiometry, R6hler et al. (1997a,b) used the Y atom for photoexcitation, due to its excellent view of the Cu202 superconducting planes. This is because, as we recall, the Y layer is situated between two adjacent Cu202 planes (figs. 4 and 5). Further, according to previous diffraction studies the Y site moves much less than the Ba site upon changing the oxygen nonstoichiometry (table 9, sect. 5.1.1). In contrast, Cu-EXAFS has the great disadvantage of two nonequivalent copper sites which makes the analysis of the spectra much more complicated. Y-EXAFS has only the drawback that it does not face the very interesting apex site which reacts sensitively to the charge transfer between chains and planes. Figure 68 shows the scattering atomic cluster with diameter~ 16A centered at the Y atom. The single scattering paths Y-O2,O3 to the nearest neighbors, and Y-Cu2 to the next-nearest neighbors, as well as the multiple scattering paths Y - C u 2 - B a and Y - O 2 , O 3 - B a are shown with arrows. The apical oxygen (tip of the pyramid) does not belong to this twoshell cluster, but it is shown for clarity. The important single backscattering (SS) and multiple backscattering (MS) paths are: Y-O2, Y - O 3 (SS) - Y-Cu2 (SS) - Y - C u 2 - B a (SS, MS) - Y - O 2 - B a , Y - O 3 - B a (SS, MS) - Y - O 2 - C u 2 , Y - O 3 - C u 2 (MS) - Y - O 2 - O 3 (MS) In the multiscattering paths, the smaller the angle between the two legs the more effective is the signal. Figure 69 shows the Yttrium K EXAFS of 123-Ox at T = 90 K, with parameter the oxygen stoichiometry (6.806 ~ x ~< 6.984). The moduli of the Fourier transforms IFT0(K2)I for various stoichiometries, as obtained from the interference patterns in the photoelectron momentum window t¢ = 2-16 ~-1, are plotted together. Xtc is the Fourier transform of an effective radial PDF p(R) which contains the structural information and the physics of electron scattering: p(R)dr ~ probability of finding an atom within a sphere with radius R + dR around the absorbing Y atom. Reff is the effective path length, R = r t o t / 2 with rtot the total length of the MS path. From the depicted Reff range the 5 strongest peaks have been analyzed in detail. As we discussed in sect. 2 (fig. 7) the CuO2 superconducting planes are not flat but have an appreciable dimpling, clearly shown in fig. 70. On average, the planes consisting of the 0 2 / 0 3 (110) rows lie 0.27 A higher along the c-axis than the Cu2 planes. In addition, the oxygen rows 0 2 and 03 lie on two different levels, and have different thermal vibrations, both resulting from the orthorhombic symmetry of the superconducting 123-Ox structure. In the unit cell the 0 2 , 0 3 layers are nearest to the Y layer and the Cu2 layers are nearest to the Ba layer. This separation between the oxygen and copper layers results in 0.8 A difference in the length of the average Y-O2/O3 and Y-Cu2 bond lengths.

122

E. KALDIS

0 2 0 0

oB

I

1 •

,

.

Reff(~) 3 4 5

2 ,

.

,

.

i

o2

,

•

,

6 •

7

,

.

,

.

Cu2

• ,oo

Fig. 68. Y-EXAFS. Scattering atomic cluster centered at the Y atom with diameter -16~,. It includes parts of two adjacent CuO 2 layers separated by the Y layer, as well as the Y-Ba block. The CuO 5 pyramid with the apical oxygen is shown for clarity. The single scattering paths to the nearest neighbors, 02,03, and next-nearest neighbors, Cu2, as well as the multiscattering paths to Ba are shown schematically by arrows. After R6hler et al. (1997a,b).

2o u_ I-'-

•

'

I

.

.

.

.

I

.

.

.

.

I

.

.

.

.

I

x

6.984 6.886-~ ~\ 0~ " ' " 6.968~ 6.947

50

6.826

5.0

5.5

6.0

Re.f(/~)

6.5

Fig. 69. (a) Fourier transforms from the Y K EXAFS interference patterns (using the photoelectron momentum window 1¢= 2-16 A ) with parameter the oxygen stoichiometry at T = 90 K. (b) Magnification of the "6.2" and "'5" peaks. After R6hler et al. (1999a).

Fig. 70. Top-side view along the Cu2 Cu2 (110) rows of the Cu202 superconducting planes of the 123-Q structure, schematically illustrating their dimpling. The copper rows lie lower than the oxygens. Inside the 02,03 rows there is a difference in height of tile 02 (lower) and 03 (higher) sites. It shows clearly that the superconducting planes consist of three stacked layers (c£ figs. 7 and 15). T h e r e f o r e , t h e i r SS s i g n a l s are clearly s e p a r a t e d in the h i g h - r e s o l u t i o n E X A F S s p e c t r a o f fig. 69. D i f f r a c t i o n w o r k h a s s h o w n t h a t t h e Y - O 2 / O 3 a v e r a g e b o n d l e n g t h is 2.40 a n d d e p e n d s f e e b l y o n t h e t e m p e r a t u r e b u t s t r o n g l y o n the o x y g e n c o n t e n t ( C a p p o n i et al. 1987, F r a n c o i s et al. 1988, J o r g e n s e n et al. 1990a, S c h w e i s s et al. 1994). C h a n g e o f s t o i c h i o m e t r y f r o m 6.6 to 6.97 leads, at 5 K, to a c o n t r a c t i o n o f the a v e r a g e Y - O 2 / O 3

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN Y B a 2 C u 3 0 , .0003

.0002

'

'

~ I

'

'

'

'

I"

'

'

'

I

'

'

'

,

I-"

,

,

'

I

x=6.968

(a)

Y-02,3

/

,

,

,

,i

~.oooI

(b) •

Ze

u

.0000

_.00010

1

-i

/"

t*)

123

"i':-i. eee

0

•

e~e#~e#e ,,,E . . . . , . . . . , . . . . , . . . . , . . . . L 50 100 150 2 0 0 2 5 0 3 0 0 80 T e m p e r a t u r e (K)

,

J

,

,

,

,

go 100 110 T e m p e r a t u r e (K)

a

-1 120

Fig. 71. Mean cubic relative displacements Ac3 vs. temperature of the highly overdoped sample x=6.968, relative to T = 100 K, to quantify the anharmonicityof the Y 02,03 bonds near Tc. (a) and (b): two different scales. After R6hler et al. (1997a). by 0.04it (Hewat et al. 1991, Conder et al. 1994a) and A ( Y - O 2 ) - A ( Y - O 3 ) ~ 0 . 0 2 A (Cava et al. 1990) (fig. 19). The two nearest- and next-nearest-neighbor peaks "O2,O3" and "Cu2" could be analyzed quantitatively. Fitting of the Y-O2,O3 peak was not possible using harmonic approximations, due to either strong disorder and~or strong anharmonicity. The anharmonicity of the radial distribution function of the Y-O2,O3 was expressed by the mean cubic relative displacements Ac3 which describe deviations from the Gaussian radial distribution function. Their plot vs. temperature shows a clear singularity at Tc (fig. 71). The mean-squared relative displacement due to small harmonic disorder of thermal/static origin, a 2 = (R-R0) 2, was found to be independent of the oxygen content in the range 6.801-6.947 but decreased appreciably (20%) for high overdoping, 6.984 (R6hler et al. 1997a). The Y-Cu2 peak does not show such effects. The crystallographic structure gives an Y-Cu2 bond length o f - 2 . 2 A and no orthorhombic split. Neutron resonance spectroscopy has shown the absence of anomalous copper vibrations in the range 10-300 K for x ~ 6.2 and x ~ 7 (Hecker et al. 1994). The pair distribution function from a harmonic fit has a nearly Gaussian form. The bond lengths Y-Cu(2) are in the range of 20-l OOK independent of oxygen content and temperature. Nevertheless, the temperature dependence of the mean-squared deviations a2(T) shows a deviation from the harmonic contribution to the Debye-Waller factor, with a clear step at 80 K and onset at Tc (fig. 72b). The peaks at 3.5, 5 and 6.2 ~t are mixtures of single and multiscattering paths and due to their complexity they could be analyzed only qualitatively (R6hler et al. 1998): Peak "3.5": Y-Ba, Y - O 2 - O 3 , Y-O2-Cu2, Y - O 3 - C u 2 , Y-Y, Y - Y Peak "5": Y-Ba, Y - O 2 - B a , Y - O 2 - B a - O 2 , Y-Ba, Y - O 3 - B a , Y - O 3 - B a - O 3 , Y - Y Peak "6.2": Y-Ba, Y - C u 2 - B a , Y - C u 2 - B a - C u 2

124

E. KALDIS '''

i ....

0.006

i ....

i ....

i ....

X=6.968 Y-Cu2

i ....

I

(a)

x10-4

(b) %/

./

~"0.004

25

/ 0.002

+] •

. . .50 ...

°'°°°o

. . 150 . . . . . . . 200 . . . . . . . . 250 ... 300 20

T e m p e r a t u r e (K)

I

40

,

l

60

,

I

80

,~

l

100

2O 120

T e m p e r a t u r e (K)

Fig. 72. Mean-squared displacements of Y-Cu2 0-2 vs. temperature: (a) the line is a fit of the harmonic model of lattice vibrations; (b) magnification of the superconducting regime. The line is a guide to the eye. After R6hler et al. (1997a). 80

,~

i

!

!

i

j-e -~P-#~

60

i

!

• 6.984 * 6.947 * 6.968

~.~-~

O.1

40

°6.806

Ii

20

0

I

0

,i

i

I

I

1O0 200 Temperature (K)

f

300

Fig. 73. Temperaturedependence of the Fourier transform modulus at the maximum of the "6.2" peak, with as parameter the oxygen content. The underdoped sample shows two steps at 160K and 100K. After R6hler et al. (1997a).

Figure 73 shows the temperature dependence o f the Fourier transform modulus at the m a x i m u m o f the "6.2" peak, with oxygen stoichiometry as parameter. There is no appreciable dependence on the oxygen nonstoichiometry except for the relatively underdoped sample (6.806) which shows a weaker temperature dependence ( - 1 5 % ) with a broad m a x i m u m at 40 K and forms a staircase with steps at T ~ 160 K and nearly 100 K. A comparison o f the scattering paths o f the "6.2" and "5" peaks reveals appreciable similarities because they both contain Y - B a configurations, but in peak "5" the intermediate atoms are the orthorhombically split 0 2 and 0 3 with shorter average scattering length and larger bridging angle. Peak "5" is nearly independent ofnonstoichiometry in the temperature range 8 0 - 3 0 0 K. This indicates that the average position o f the oxygen layer 0 2 / 0 3 is nearly independent for x = 6.801-6.984. The heavily overdoped sample (6.984) shows a higher signal for T < 8 0 K (10% at 40K).

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN YBa~Cu30x

125

The temperature and oxygen concentration dependencies of the maxima of peaks "3.5" and "6.2" are highly correlated. They both have a relatively weaker, staircase-like temperature dependence of the underdoped sample and a similar temperature dependence of the overdoped samples at T > 80 K. A difference is that the overdoped samples of"3.5" show a well-developed plateau in the 100-130K range and a clear step at 100K. Model calculations neglecting vibrational displacements by R6hler et al. (1998) could qualitatively reproduce the Fourier transform spectra found experimentally. The main assumption has been that with changing oxygen nonstoichiometry Cu2 moves along the c-axis, neglecting vibrational displacements. This work allowed to understand better the influence of the Cu2 shifts on the spectra. The main results emanating from these EXAFS investigations (R6hler et al. 1997a,b, 1998) are: - Y-Cu2 spacing. The experiment shows that the Y-Cu2 bonds are independent of oxygen nonstoichiometry. On the other hand, the model calculations of R6hler et al. (1998) show that with increasing Cu2-O2,O3 spacing (increasing oxygen content) peak "5" remains unaffected, the heights of peaks "3.5" and "6" increase (both in good agreement with the experiment) but the position of the "Cu2" peak is also shifted. The latter seems to contradict the experiment. This discrepancy is solved if it is assumed that the shift of Cu2 along the c-axis takes place under constant Y-Cu2 bond length. Under this condition the Y-Cu2 separation (along the c-axis) increases with oxygen content only if the Cu2 atoms are allowed to rotate around the central Y sites in the [111] plane o f the Y-Ba blocks. This results in an expansion/contraction of the cationic sublattices of Y, Cu2, and Ba along [110], in agreement with the contraction of the a-parameter and the expansion of the b-parameter with increasing oxygen content found by X-ray diffraction of equilibrium samples (Conder et al. 1994a, KNger et al. 1997, Kaldis et al. 1997a), discussed in sect. 3.2.2.3. Figure 74a schematically illustrates these distortions with exaggerated orthorhombicity. The smaller a/larger b configuration corresponds to the optimally doped a - b plane, and the opposite to the underdoped. The arrows show the three-body scattering configuration Y - C u 2 - B a of the "6.2" peak. A side view of the [110] crosssection along the c-axis, fig. 74b demonstrates how under constant Y-Cu2 bond length the separation between Y layer and Cu2 layer increases by 6z and the scattering angle decreases by 2 °. With doping the orthorhombicity increases and the rotation around Y shifts Cu2 lower along the c-axis, thus increasing the Y-Cu2 separation. Further conclusions about the Y-Cu2 pair are that in the normal state they vibrate harmonically, but in the superconducting state their vibrations are frozen. Symmetrybreaking lattice distortions do not exist in the Cu2 sublattice. - Y-O2,O3 spacing. Its dependence on nonstoichiometry is controlled from the two conditions: Y-Cu2 bond length constant (see above) and peak "5" practically independent of nonstoichiometry (experiment). As already mentioned, this peak has very strong contributions from the Y-O2,O3-Ba MS paths. From Figure 75a it can be seen that the changes of the Y - B a block with nonstoichiometry influence these scattering paths. Varying the Y-O2,O3 distances inside the rather large error margins

126

E. KALDIS

~//

(b) [110] -

Y

•

--,

R(Y-Cu2)

Ba

Cu2

" ~

~

'qv

Fig. 74. (a) Top view of the Y (upper) Ba (lowest) layers with Cu2 and 02,03 between them (Y, small open spheres; Ba, large black spheres; O, large open spheres). The geometry of the tmderdoped and optimally doped cells of 123-O x are overplotted. The increase of the orthorhombicity with oxygen doping is exaggerated (small alarge b-axis optimally doped). The Y-Cu2-Ba scattering paths (peak "6.2") are indicated by arrows. (b) Crosssection along [110] with constant Y Cu2 bond lengths to illustrate the rotation of Cu2 with increasing doping, leading to an increase of the Y-Cu2 separation by 6z. After R6hler et al. (1997a).

a

b

O2 b. Fig. 75. (a) The same cell as in fig. 74a, rotated by 45 °. The arrows show here the Y-O2,O3-Ba scattering paths (peak "5"). (b) Analogous cross-section as in fig. 74b, along [010] with the Y-O2,O3 Cu2 forward scattering angle fixed at N13° on doping. As can be seen, doping does not shift the oxygens along the c-axis (6z ~ 0) but along the a- (cf. a) and b-axes, the Y 02,03 bond lengths changing in the frame of the orthorhombic distortion. The dashed horizontal lines show the expansion of Cu2 along the c-axis. After R6hler et al. (1997a).

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTS1N YBa2Cu30~

127

(~1%) of the harmonic single-shell analysis of the " 0 2 , 0 3 " peak, and restricting the 0 2 , 0 3 displacements only along the c-axis, two possibilities may be considered: fixed forward scattering angle or fixed average Y-O2,O3 distance. In the first case the O2,O3 displacements along c are very small as shown in fig. 75b. In the second case these displacements and, therefore, the changes of the forward scattering angle become appreciable (fig. 75b), shifting the height of the "5" peak outside the experimental findings. The 0 2 and 03 move along the a- and b-axes according to the orthorbombic distortion. We note the strong anharmonicity and/or disorder of the Y-O2,O3 bonds mentioned above. In addition, the shapes of the " 0 2 , 0 3 " and "5" peaks in the calculated Fourier transform spectra have some differences to those found experimentally. This indicates also that the ideal geometry of the Y-02, 03 cluster used in the calculations is not correct and that this cluster is distorted. Finally, Rrhler concludes from this analysis that, contrary to the Cu2 sublattice, symmetry-breaking distortions (e.g., of the antiferro type) would be allowed in the 02,03 layer. - Cu-O2,O3 spacing. As we have seen, peak "5" contains nearly collinear Y-O2,O3-Ba multiscattering paths (deviation 13°) and peak "6.2" even more collinear Y - C u - B a (deviation 4°). Therefore, a very small variation of the forward scattering geometry will have very strong influence on the effective amplitudes (R6hler and Crfisemann 1995). Calculations with the FEFF6 code (Zabinsky et al. 1995) show that the maximum of the "6.2" peak varies almost linearly with the c-displacements of Cu2 (Rrhler et al. 1997b). In view of the fact that both scattering paths involve the same Ba layer, the dependence on nonstoichiometry and temperature can be taken from the ratio of the Fourier transforms of these two peaks. The results are shown in the next section. As discussed above, the peaks "5" and "6.2" have quite different dependencies on the oxygen nonstoichiometry. The ahnost negligible change of peak "5" for x < 6.95 indicates that the vertical position of O2,O3 is not influenced by the nonstoichiometry in the underdoped and optimally doped regime. The increase of peak "6" shows that the Cu2 atoms do shift appreciably towards the Ba layer, moving out of the plane. The overall increase of this peak corresponds to a relative displacement of Cu2 by

6R=0.051~. 6.5. The displacive martensitic transformation in the ooerdoped regime of 123-0x The use of equilibrium samples (cf. sect. 3.1.2) and the high resolution of the volumetric oxygen analysis (sect. 3.1.3) allowed a systematic investigation of the optimally doped and overdoped regime of 123-Ox up to x = 6.984. We recall briefly the steps discussed in previous sections which led to the discovery of the displacive martensitic transformation. A minimum of the c-axis in the oxygen-rich region x = 6.925 (fig. 28a, sect. 3.2.2.4), found only for the BAO and DO equilibrium samples by XRD (sect. 3.1.2), was confirmed by ND (fig. 290 and was found to result from an abrupt discontinuity of the dimpling (Cu2-O2/O3) of the superconducting planes (fig. 66a) indicating the existence of a phase transition (sect. 6.2). Micro-Raman investigations (statistical average of many crystallites)

128

E. KALDIS

0,30 • ~

-'

6,80 6,85 underdoped

'

t

.

.

.

.

I

6,90

'

•

• -'

0,29-

123-Ox EXAFS, T=25K C~} 0,28- R0hler et al.1997a,b Kaldis et al. 1997 ~

0,26 0,25

.......-'"""

y

0,27-

'

I

.

6,80

.

.

.

6,95 .

optimally' J doped displacive martensiti¢ transsformatJon

...

7,00 • . 0,30 pverdoped t -] 0,29 i~.. ~

t0'28

i~

J 0,27

.L, I . . . - . . 6,85 6,90 6,95 7,00 oxygen content x I

.

.

.

.

. 0,25

Fig. 76. Dimpling O2/O3-Cn2 as a function o f x (after R6hler et al. 1997a,b and Kaldis et al. 1997b). Note the abrupt first-order jump at the onset of the overdoped regime, x = 6.947, which has been completely confirmed by later measurements (R6hler et al. 1998). Note that without the two points in the range x = 6.95-6.97 the b'ansition step would have been missed.

showed that one of the phonons corresponding to dimpling, the Ag phonon of the in-phase vibrations of 02 and 03 (J[ c-axis), showed a strong softening at x=6.950 (figs. 65, 79) (sect. 6.2, Poulakis et al. 1996). EXAFS found a discontinuity there (R6hler et al. 1997a,b) (sect. 6.4), and finally when 10 samples with different oxygen contents could be measured (R6hler et al. 1998), a steep first-order-like jump appeared in the dimpling (fig. 76). These last measurements have been performed recently at four additional compositions, three indicated on the x-axis and one at 6.95. They completely support the above figure. From fig. 77 we see the structural changes leading to this transformation. With increasing oxygen content the Y-Cu2 separation increases (with the Cu2 moving on a sphere of constant Y-Cu2 radius, cf. fig. 74) by the shift of Cu2 towards the Ba layer. For x < 6.950 the positions of 0 2 and 03 remain practically unaffected. It is in the overdoped regime that the 02, 03 move away from Y. As in this range the a-axis increases, and the b-axis decreases (figs. 25a,b) increasing the tetragonality of the unit cell, this transition is considered to be a martensitic transformation of 123-Ox (Kaldis et al. 1997b). Closer inspection of the spikes of the lattice parameters vs. x of the DO samples (sect. 3.2.2.4), in particular those of the c-axis or of the unit cell volume, shows a striking similarity to the changes of dimpling shown in fig. 76, for CAR samples. Figure 30d illustrates this unexpected behavior of the c-axis at larger magnification. The minimum of the DO c-axis corresponds to the optimally doped phase. The step at x = 6.95 corresponds to the onset of the overdoped phase and the martensitic transformation. There is a sharp spike at x = 6.970 corresponding to the 0 2 , 0 3 shift in the opposite direction (away from yttrium) (fig. 77), strongly decreasing the dimpling. Unfortunately, no data point exists in the range 6.970 < x < 6.980 to judge the slope of the dimpling in this range. One could

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30x

~

1,74 I 1,731

.-- 1,70 ~ 0 Q. 1,69 co

T

129

optimally doped

underdoped EXAFS 123-O, T = 25+5 K R6hler et al. 1997a,b Y-Cu2

A

/i

y

-

ee~

q~2 g =~

O

9

B~a,Cul"

1.45 ~"

1,68 ~ 1,67 1,66 6,80

.u-..... 6,85

_ ~ 6,90

-~

Y-02

6,95

3

1.43

~~.

1.42 7,00

Or)

.

Oxygen content x Fig. 77. Variation of the spacings Y-Cu2 and Y 0 2 , 0 3 with oxygen nonstoichiometry. Note the abrupt jump of the average 0 2 , 0 3 position towards Cu2. In the underdoped regime, increasing distance of 0 2 , 0 3 from Cu2 scales with increasing To. After R6hler et al. (1997a,b, 1998) and Kaldis et al. (1997b). On the right, the movements of the atoms along the c-axis are shown schematically. The Cu2 atoms move out of the superconducting plane.

argue that the position of the spikes (fig. 30d) in the lattice parameters of the DO samples is an extreme coincidence. Some arguments against this were presented in sect. 3.2.2.4. The transition at x = 6.95 underlines the change of mechanism from the optimally doped superconductor to the "bad" metal in the overdoped region. An important characteristic of this transition is its influence on the magnetization. As we will discuss in sect. 7.1 a splitting of the diamagnetic transition has been observed by many laboratories for oxygenrich samples. For the equilibrium samples discussed here this splitting appears without exception at x = 6.949, exactly at the doping leading to the martensitic transformation at the onset of the overdoped phase (Kaldis et al. 1997b). Thus, this transition can be considered as the onset boundary of the phase-separated ouerdopedphase where the Tc decreases with nonstoichiometry. Recently, Emerson et al. (1999) have confirmed with careful specific-heat measurements previous findings of Nakazawa and Ishikawa (1989) and Moler et al. (1997) that in the overdoped regime a sharp increase of the Debye temperature takes place. This is shown in fig. 78. In their estimated oxygen content scale the abrupt increase of the Debye temperature takes place at x ~ 6.98. In view of the fact that the oxygen determination was indirect, it seems reasonable to assume that this effect is related to the displacive martensitic transformation (x = 6.95, determined directly, with high accuracy, sect. 3.1.3) and the series of structural phenomena which accompany it in the overdoped region: the abrupt decrease of the unit-cell volume (fig. 25e), the decrease of the b-axis, the decrease of the orthorhombicity (figs. 25b,c) and the minimum of the c-axis (Rusiecki et al. 1990) (fig. 28a).

130

E. KALDIS

pp

19g,8° P,,~5

p,9~ ~,00 ~'ff95

123-0, ,.-,~'190- Emerson etal. .=

i

185-

E

190

185 180

18o-

j~ 175-

175 J

Gl 17o , 6,80 .

.

.

6,85

.

i

.

.

.

.

6,90 .

.

.

.

6,95

.

.

.

.

.

.

.

170 7,05

.

7,00

oxygen content x

~

456454 452 -" ~'450448 I L 446 444442 ~ . 440 438 -

'""~i" T

"'""'io

J,

-'1~

~'-1

',,

,

"~1

~.~..

I-

~ I~'i

I-u-i,

~,~

430"

-J

J i -I

IPoulakis et al. 1996 [Palles eta,. 1999

1~.= 432 ¢"

1-1

~i._~,

436 2 434-

'

I

•

6,0

'

'

I

6,2

'

~

'

Fig. 78. Sudden increase of the Debye temperature at x ~ 6.95, from specific heat measurements. After data of Emerson et al. (1999).

I

6,4

'

'

~- ] '

I

6,6

'

"

'

I

"

6,8

Oxygen content x

'

I

7,0

Fig. 79. Raman in-phase phonon frequencyvs. x. Average values from 5-10 crystallites at each composition. Compare with fig. 65 showing the individual measurements. The onset of the abrupt softening of the phonon is clearly seen at the transition stoichiometry(x = 6.95). After Poulakis et al. (1996), Palles (2000) and Palles et al. (2000a).

Another change o f properties influencing the Debye temperature is the abovementioned softening o f the Raman in-phase phonon, already at room temperature, accompanying this transition (figs. 79, 65) and the independence o f the apex phonon (fig. 61) in the overdoped regime. We note also that these structural changes may have important implications for the physics o f HTc (R6hler 2000a). According to Andersen et al. (1995) the shifts o f Cu2 and 0 2 , 0 3 out o f the plane influence the band structure in the vicinity of the Fermi level.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30 )

131

7. Overdoped phase - II: phase separation in 123-0x

7.1. The splitting of the diamagnetic transition in the ouerdoped regime Although theoretical models of phase separation due to the separation of spin and charge appeared very early (Gor'kov and Sokol 1987), the research concentrated on La cuprates, and only very recently has appreciable experimental evidence appeared about the existence of stripes in 123-Ox. The phase-separation literature has been reviewed in a series of volumes edited by Mfiller and Benedek (1992), Sigmund and Mtiller (1994), Mihailovic et al. (1995), Kaldis et al. (1997a), and Bianconi and Saini (1997, 2000). Most of this work is based on the proximity of the superconductor and antiferromagnetic insulator states in the HTc cuprates. A system of doped holes in an antiferromagnetic insulator matrix tends to phase-separate. Historically, the first indication from materials research for the existence of an overdoped regime came with the finding of a maximum of Tc (Graf et al. 1990, Rusiecki et al. 1990) and with the c-axis anomalies discussed in sect. 3.2.2.4 (Rusiecki et al. 1990). More data were presented later by Claus et al. (1992a,b). However, in the earlier literature there is a series of indications for several anomalies of the oxygen-rich samples. Splitting of the specific-heat peak for oxygen-rich samples has been observed in longannealed unquenched samples as early as the Interlaken Conference, by Ishikawa et al. (1988a,b). Figure 80 shows their first results with two transitions at T ~ 88.5 K and 91.3 K. A similar splitting was found in the resistive and magnetic transitions, leading them to the conclusion that both peaks correspond to superconducting transitions involving substantial fractions of the sample volume. Further results of this group will be discussed below. Also, work on single crystals by other groups gave similar results (Inderhees et al. 1988, Butera 1988). Phase separation in two orthorhombic phases with x ~ 7.0 and 6.7 but common orientation was found with XRD high-resolution scans in a triple-axis spectrometer using Ge (111) crystals as monochromator and analyzer (YouHoydoo et al. 1988). '

n

,

n

'

K

,

140-

i

'

~

'

~

'

n

'

u

'

n

,

':

,..-,..

"O

mmn mammmSmmmnm

,E~ 1 3 6 -

I,--

134-

"~ (,,)

1321 3 0 Im,

82

mmu.mN i""i"" 1 2 3 k a w a et al 1988

f .

.

,

8'4 8'6 88

.

,

90

.

.

,

9'2 9'4 96

Temperature (K)

.

,

.

.

96 160 102

Fig. 80. Specific heat Cp/T as a function of temperature for an oxygen-rich sample with two distinct peaks (ACp/T= 3 K). This splitting was found also in the resistive and magnetic transitions. The figure was schematically reconsmacted, not considering all measured data. After Ishikawa et al. (1988a).

132

E. KALDIS

Loram et al. (1991) investigated the specific heat of oxygen-rich samples as a function of nonstoichiometry. Using a high-precision differential technique, with as reference a YBa2(Cu0.93Zn0.7)3Ox sample to suppress the large phonon term, they found that their highly oxygenated 123-Ox (x ~ 7) samples showed a double transition in the range of x ~> 6.95. The splitting changed systematically with x and was present in all samples. At x ~ 7 two peaks were observed at Tel ~ 88 K and Tc2 ~ 92.5 K in 4 samples from three different laboratories. Resistivity measurements showed similar effects. These authors also considered the specific-heat anomaly to be a bulk effect and to be definitive evidence for a multicomponent order parameter. Several arguments support a bulk effect (Loram et al. 1991): The comparable magnitudes of the anomalies at Tel and Tc2 show that each of them is connected with approximately half the sample volume. The authors explicitly reject the existence of minority (impurity) phases with a different Tc than the bulk, or nucleation of superconductivity on grain or twin boundaries. The sharp anomalies are not consistent with a spread of T~ due to inhomogeneities in the oxygen distribution and with the sequence in which the measurements were performed (first x ~ 7.0, then 6.50, then 6.75). - As the resistivity onset of the diamagnetic transition coincides with To2 it is clear that the long-range order of superconductivity starts at this temperature. As the resistive transition is finished at T > T~ the authors conclude that the long-range order is fully established at this temperature. As a possible explanation of this phenomenon the authors consider, in the case of non s-wave pairing, the coupling between the superconducting order parameter with an environment of different symmetry and propose that it may lead to two different T~'s dependent on the crystallographic direction. Thus, Td, which changes with the chain oxygen, may be the transition temperature for c-axis superconductivity while To2, which seems rather independent of oxygen, may be the transition temperature for the superconductivity of the planes. The crystallographic phase separation (chemical phase separation) (YouHoydoo et al. 1988) reappeared in the work of Claus et al. (1992b). They found that single crystals with x > 6.92 had split c-axis with slightly different c-parameters. The corresponding resistive and (dc) diamagnetic transitions show the two-step behavior of Loram et al. (1991), but their specific heat did not show the two peaks. For the structural investigation they used a two-axis goniometer and recorded 0 - 2 0 XRD spectra in the Bragg-Brentano focusing geometry. Only c-axis reflections (00l) could be measured for l = 2 - 1 4 . The absolute accuracy of the c-axis parameter was estimated to be rather low, -0.01 A. A narrow rocking curve (-0.025 FWHM) was measured for one crystal; other crystals had much broader rocking curves corresponding to misorientations of 0.5 °, and some showed several peaks indicating that they consisted of several grains with similar misorientations. However, each of these grains (-6) could be measured individually. The crystals were thermally treated under known conditions to reach oxygen contents near 6.92 and then 6.94 reversibly and were measured in the two states. Figures 81a,b show the lattice constants and the relative intensities of the crystal parts, with larger c-parameter -

-

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30x 11,72

. . . .

,

.

.

.

.

I

.

.

.

.

I

I

ing[e phase

,1

I

I

]

.

.

" i 2 phases

.

133

.

(a)

_~ 11,71! 2

11,70-

i ',

a5

',

E 11,69-

1 2 3 - O ~ ~ 4 ~, 11,68-

g'

1992

U 11,67-

C1

3

i

I

,.54i 1D

t~ 11,66-

=14

, ~ C 2 11,65 .... 6,75

, .......... . !.. L , . . . . 6,89 6,85 6,90 6,95 7,00 Oxygen content x

. . . .

r

. . . .

r . . . .

i

. . . .

i

. . . .

12-

(b) C1

lO .

""

.

.

.

'5

.3

21

8-

+~6.

J

z"

.

1230~ XRD Claus et aL 1992

0 ........... 6,75 6,so

6,85 Oxygen

=4

6,90 content

~'/ 6,95 x

7,oo

Fig. 81. (a) Splitting of the XRD c-lattice parameters for crystals reversibly oxygenated and deoxygenatedbetweenx ~ 6.92 and 6.94. At x < 6.92 the lattice constants follow the normal x dependence (C1). At x>6.92 the splitting starts (C2). (b) Relative intensities of the (0013) diffraction line of the part of the crystals with the larger c-parameter (C 0. Above oxygen content 6.94 (C1) the intensity disappears reversibly with the oxygen content. The numbers show measurements on the same crystal. After Claus et al. (1992b).

Cj at x~<6.92 and smaller C2 at x~>6.92. From fig. 81a we see a miscibility gap, two phases coexisting in the 6.92-6.95 range. From fig. 81b we see that the intensity o f the larger (C1) disappears - as expected - with increasing x, after we cross the miscibility gap. This is the classical picture o f phase rule thermodynamics. In view o f the estimated oxygen contents they used, it is possible that these figures illustrate the c-parameter anomaly (fig. 28a) and the displacive phase transformation found at the onset o f the overdoped region at x = 6 . 9 5 (Rusiecki et al. 1990, Conder et al. 1994a, Kaldis 1997, Kaldis et al. 1997b, R6hler et al. 1998), which also lead to phase separation. Possibly due to hysteresis, the grains o f the various crystallites measured above could be frozen at different stages o f the transition and the minimum o f the c-parameter could show, therefore, different dimplings and c-parameters. Figures 81a,b also remind o f the miscibility gaps tentatively proposed for the a-parameter o f the BAO samples (fig. 29c). 7.2. The shell model Soon after the work o f Loram et al. (1991) and before the work o f Claus et al. (1992a,b) Nakazawa finished his Ph.D. thesis (Nakazawa 1991) at Tokyo University. This thesis

134

E. KALDIS

summarized and discussed the numerous investigations performed since 1987 by the Ishikawa group, with particular care for the quality of the material (e.g., Ishikawa et al. 1988a,b). The highlights of this thesis were published later (Nakazawa et al. 1994). These authors were, therefore, the first to show that improvement of the sample preparation by avoiding quenching and by the use of low-temperature annealing allows the development of the two specific-heat peaks at Tc (fig. 80), and were also among the few teams at that time which measured directly the oxygen content of each of their samples by iodometry (Nakazawa 1991, Nakazawa et al. 1994). They also developed the shell model we will discuss below (Nakazawa 1991, Nakazawa and Ishikawa 1991). An extensive investigation of the specific heat and the a.c. susceptibility as a function of the annealing temperature ofpolycrystalline samples was presented by Janod et al. (1993). They found for all investigated samples the double peaks for annealing temperatures 350-550°C, except for 460°C. It is unfortunate that their expertise in specific-heat and ac-susceptibility measurements could not be supported by accurate determination of the oxygen content of each sample. They did not measure the oxygen of the samples they used for specific-heat and susceptibility measurements, but that of another series of samples which showed a smaller splitting. Thus, all the figures of this work have only the annealing temperatures as parameters, and only the Tc vs. x shows estimated oxygen contents based on the work of Claus et al. (1992a), which in turn also used thermodynamically estimated oxygen contents. Further, no investigation of the change of the lattice constants with oxygen was presented, which would trace the trends of structural changes. Nevertheless, the work of Janod et al. (1993) is an important contribution to the splitting of the diamagnetic transition because it supports and extends the experimental findings of Ishikawa et al. (1988a,b), Nakazawa (1991) and Loram et al. (1991): (a) that careful straightforward synthesis with low-temperature annealings leads to two very well-developed peaks at To, and (b) that both peaks correspond to superconducting transitions and to substantial superconducting volumes of the sample, reversible with oxygen change. The two peaks of the specific heat of 5 samples annealed at 300°C for 240 h appear at two critical temperatures Tel = 87 K and To2 = 92 K, corresponding to 40% and 60% of the sample volume. With increasing annealing temperature a cross-over of Tel and To2 takes place in the samples of Janod et al., as shown in fig. 82, but unfortunately we do not know exactly the corresponding change of the oxygen or carrier concentration. The authors claim that they have reached an oxygen content of x =7.0 under 100 bar of oxygen and refer to a Russian team who have reached x = 7.2. It is questionable whether such samples are homogeneous at mesoscopic scale, due to the T - x phase diagram shown in fig. 55 (Karpinski et al. 1991) and discussed in sect. 5.4. Also one would like to know more about the accuracy of the indirect methods used for such oxygen determinations. Janod et al. (1993) propose the same model as Nakazawa (1991), namely that the grains of the sample consist of a surface layer (shell) with higher oxygen content and lower Tc, and a core with lower oxygen content and higher Tc. Nakazawa's model is based on the lattice-constant anomalies reported earlier (Rusiecki et al. 1990) for x=6.92, which he could reproduce in carefully synthesized low-temperature-annealed samples: a minimum

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30,

94

94

core (O-poorer) 92-

~

135

,o00)

2)

~6.91

/

,

92

\

'90

90(6.8

///~

I-- 8s(720)

123-0× Janod et al. 1993

86300

surface (O-richer)

3;0 400 4;0 6()0 annealing temperature [°C]

~

88 ~

86 550

Fig. 82. Evolution of Tel and T.2 with annealing temperature. After data of Janod et al. (1993). The authors propose that T¢1 corresponds to an oxygen-rich shell on the surface of the grains and To2 to an oxygen-poor core of the grains.

of the c-parameter and a maximum of the orthorhombicity (Nakazawa et al. 1994). Based on these structural changes Nakazawa proposes that in the surface layer of the grains with x > 6.90, 05 sites are occupied, which decreases To. In fact, the same argument has been used later by Krfiger et al. (1997) to explain the changes of the lattice constants in the overdoped regime, but this could not be proven experimentally up to now (sect. 5.4.2). However, in order that the Nakazawa/Ishikawa/Janod et al. shell model is valid, the surface layer has to strongly reduce the diffusion of the oxygen through the surface to the core of the grains. As we will discuss in the next section, this is not the case (Conder et al. 1994a,b). 7.2.1. Oxygen diffusion in the grains: no surface barrier In order to test the above "shell" model, the diffusion rate of oxygen in 123-Ox grains at 320°C was calculated for various grain sizes, using measured diffusion coefficients (Conder et al. 1994b). Self-diffusion coefficients were measured in thermogravimetric studies of the oxygen isotope exchange (Conder et al. 1993). The chemical diffusion coefficient D was calculated from the self-diffusion coefficient D* using the equation D = D*Fh,

(12)

with an estimated thermodynamic factor F and the Haven factor h which has a value near unity (Salomons and de Fontaine 1990). Due to the strong anisotropy of the oxygen diffusion coefficient in 123-Ox a onedimensional bilateral diffusion model (Schmalzried 1981) was used for the calculation of D (Conder et al. 1993). From the solution of Fick's second law the total concentration of the diffusing oxygen was obtained. This is equivalent to the diffusion progress a, which is 0 at the beginning and 1 at the end of the diffusion process. This parameter has been determined directly from thermogravimetric weight-change measurements (Conder et al. 1993, Kr/iger et al. 1993) of the oxygen isotope exchange 180 ~ 160 at very low temperatures (225-290°C, thermobalance resolution 1 ~tg). The experimental results

136

E. KALDIS 800 1E-10

750 •

.

I

700 "

I

650 "

!

600 •

550

I

"

'"

500

!

"

[K]

I

1E-11 "~"

E.=O.97Z0.05 eV " ' . "...

~

E==1.23-'0.15 eV

1E-12

a 1E-13

r,

single crystal= (RothmanetaL 1 9 9 1 / polyc.,,~tals (Rothman et al. 1989)

•

1E-14

~

t=~,, x ,, " ~

P°wderlS-501am'%(Conderetal. 1994b)

a powdernot sieved" ,

1.3

1;4

l'.s

,

1.6

13

1;8

,

1.9

, ......

i.o

1000/-r [K4] Fig. 83. Arrhenius plot of the oxygen tracer self-diffusion coefficient D*. Comparison of the data of Conder et al. (1994b) with those of Rothman et al. (1989, 1991). After Conder et al. (1994b).

showed that at these temperatures only the 0 4 sites of the chains are exchanged, justifying the use of a one-dimensional model (Conder et al. 1993). As in these experiments the oxygen content remains constant, the tracer-diffusion (self-diffusion) coefficients can be calculated from isothermal runs. To receive as high an oxygen content as possible, the synthesis method of sect. 3.2.1 (table 2, CAR method) was extended by an additional annealing at 320°C for 300 h in 1 bar of oxygen or in other samples by slow cooling down to 500°C with 5°C/h and then by 4"C/h down to room temperature. The oxygen tracer coefficient D* was calculated for many values of a obtained from the isotope-exchange experiments at different temperatures. Satisfactory constant values near 10 13 cm2/s were obtained in the vicinity of 250°C. Arrhenius plots are shown in fig. 83 including the data of Rothman et al. (1989, 1991). We note that their data have been based on secondary ion mass spectroscopy (SIMS), taking depth profiles of the tracer in the sample. Conder et al. (1994b) from their high-resolution thermogravimetric measurements obtained the value E a = 1.23-4-0.15eV at low temperatures for powder samples with grain size 15--50 ~m. To calculate the chemical diffusion coefficient from eq. (12), the thermodynamic factor F must be known. F is given by (Murch 1980, Salomons and de Fontaine 1990) F=0.5

d ln(po/p° ) d ln(7 - x) '

(13)

with Po2 the oxygen pressure and 7 - x the oxygen content of the sample. Therefore, F can be calculated from the slope of the 123 equilibrium isotherms in the ln(po2)l n ( 7 - x ) coordinates measured earlier (Conder et al. 1992). Using the value F = 2 0

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN YBa2C%Ox

137

0.0

,

•

..., oeO~" oOxo@ Ox@

o°xe

d < 15urn

o

XO O

X •

×0 °

-0.2

•

d

>

50urn

X °°

x o

unfiltered

-0.4

-0.6

-0.8

, 85

87

89

, 91

, 93

T (K)

Fig. 84. Magnetizationcurves vs. T for three samples, two of them resulting from sieved fractions of the first, after grinding. Solid lines are Shoenberg fits (cf. sect. 7.3). (a) Equilibriumsample (crosses), T~,o,~ot = 92.2 K. (b) Sieved fraction with grains <15 btm (open circles). (c) Sieved fraction with grains >50 p,m (solid circles). No changes of T~]= To,onset=92.2 K and Tc2 = 90.2 K with grinding or grain size were found (see text). After Conder et al. (1994a,b). (corresponding to x = 6.93) the chemical diffusion coefficient for 320°C was found to be D = 8 × 10-11 cm2/s. With these values, grains up to 100 ~m should be completely oxidized after annealing at 320°C for 300h. Using the data of Rothman et al. (1989, 1991) with F = 2 0 one calculates a chemical diffusion coefficient D = 1.4 × 10- j l cmZ/s. With this smaller value which does not correspond to equilibrium samples the grains remain homogeneous up to 50 ~tm. 100 gm grains would show only a difference A x = 0.02 between surface and center. To confirm these calculations directly with magnetization experiments an oxygenrich sample (To,onset= Tel = 92.2 K) was measured and then fractionated with ultra-sound sieving equipment to three fractions of different grain sizes. The magnetization of the two extreme fractions with grain size d < 15 ~tm and d > 50 btm was measured again and is shown in fig. 84. From the fit in the temperature regime T < 8 8 K (see sect. 7.3) it is found that the broadening of the transition at T > 88 K - leading to two Tc's is observed for both fractions in the same temperature range, both fractions having an identical Tc2=90.2K. The second fit parameter R/Xo (eq. 15, sect. 7.3) for the fraction d > 50 ~ m is ~3 times larger than that for d < 15 ~m, in agreement with the grain sizes.

138

E. KALDIS

This shows that both fractions have grains with identical magnetic properties. The small difference in the shape o f the magnetization curves is, therefore, due only to the different grain sizes and not to different superconducting properties of the grains. The Shoenberg equation (Shoenberg 1940) (eq. 14, sect. 7.3) fits the magnetization curves of both grain sizes (fig. 84) very well, but only in the T < 89.9 K regime. This means that the anomalous magnetization behavior near Tc is not due to an inhomogeneous oxygen distribution in the grains, because this would have been different for different grain sizes also for T < 89.9 K (Conder et al. 1994b). The above magnetic and thermodynamic investigations show that the "shell" model cannot be used to explain the splitting of the diamagnetic transition. 7.3. The splitting of the diamagnetic transition as a function of ttle exact oxygen content." indication for phase separation The magnetic and structural investigations started by Rusiecki et al. (1990) were continued in the following years and as mentioned in sect. 6 led to the discovery of the displacive martensitic transformation at the onset of the overdoped range (Conder et al. 1994a, Kaldis 1997, Kaldis et al. 1997b, R6hler et al. 1998). The magnetization of more than 150 samples analyzed with high-resolution volumetric analysis, Ax = 0.001 (Conder et al. 1989), has been measured, and some Tc values are shown in fig. 2 and were reported in several occasions (Schwer et al. 1993a, Conder et al. 1994a, Kaldis 1997). At this point we concentrate on the overdoped regime. As fig. 85 shows, the splitting o f the diamagnetic transition starts for equilibrium samples precisely at x=6.950, i.e., at the onset of the overdoped regime coinciding with the martensitic transformation (sect. 6). This 110

'

i

'

i

,

i

,

optimally doped

'

,

i

.

•

'

overdoped

100....................

•m

........

J

m

'1

~ ' 90-

•

•

m

,,, Tc2,°"set

es m

m u m

IE 123-0×, I~ Conder et al. 1994 I 80" Zech et a1.1995 I I.~ i_~ i.~_ 70 6,92 6,94 6,96

~n

Tel

e~

Q.

"1o

i

•

•

oxygen

I

,

'

i

.

'

6,98

content

,

7,00

7,02

x

Fig. 85. Splitting of the diamagnetic transition of 123-O~ at x >~6.950, indicating the phase separation of the overdoped phase. Tol = T c , optim, doped remains practically unchanged with x, whereas To2 decreases with x. The onset of the splitting coincides with the displacive martensitic transformation (sect. 6.4) at the onset of the overdoped phase. After data of Conder et al. (1994a) and Zech et al. (1995a,b).

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTS1N YBa2Cu3Q

139

splitting has been found for all investigated samples' with x ~>6.950. DC magnetization measurements have been performed with a SQUID magnetometer on cold-pressed powder samples (to avoid recrystallization of surface grains when sintering ceramic pellets) sealed under vacuum between two quartz rods in quartz ampoules (internal diameter 3 mm) (Zech et al. 1995a). High-precision field-cooled (FC) measurements were performed while cooling down from T > Tc with an external field H = 10 Oe. Measurements at very small temperature intervals (down to every 0.5 K) were taken in order to clearly expose changes of slopes. We recall that FC magnetization measures the field repulsion of the single grains, in contrast to zero-field cooling (ZFC) which measures the screening of the applied field and therefore, the superconducting properties of the entire sample surface. Figure 8 shows typical magnetization measurements for an optimally doped sample (x = 6.912) and an overdoped sample with x = 6.974. A two-parameter fit with the Shoenberg model is shown by the solid line. This simple model describes the flux expulsion of small grains with diameter comparable to the penetration depth (Shoenberg 1940) according to the equation M(T)~o~m = 1 - 3 ~(RT)-coth( ~ T ) )

- 3 (~)

2

(14)

with Mnormthe normalized magnetization, R the radius of the grains, and )~(T) the London penetration depth. For the fitting the two-fluid approximation was used: a(r)2 _

z ( T ) o2 (1 - T/T~)4'

(15)

with X0 =)~(T=0K). The only fitting parameters are Tc and R/)~o (Zech et al. 1995b). Figure 8 (curve a, x--6.912) shows an excellent fit up to To, with eq. (14), for the optimally doped samples. It also shows that for overdoped samples (curve b) an excellent fit is possible, but only up to a lower temperature where a broadening of the transition takes place due to change of curvature. For the overdoped sample with x = 6.974, eq. (14) describes the magnetization for T < 89 K. An extrapolation of the fit up to M(T) = 0 gives a Tcl = 89 K clearly different from the onset of the diamagnetic signal at Tc2 = 92.2 K. We recall that these are very near to the values found by the specific-heat measurements of Ishikawa et al. (1988a,b), Loram et al. (1991), and the crystallographic splitting (Claus et al. 1990; sect. 7.1). This effect shows the existence of two systems with different critical temperatures. In the underdoped and optimally doped phases only one transition temperature (To) is found. As fig. 85 indicates, the Tc2 values fit without discontinuity to the optimally doped To values. The lower T~ values in the overdoped range decrease with x so that, e.g., AT~ ~ 3.5 K at x = 6.976. More details about the splitting can be seen in fig. 86, showing the susceptibility curve of a x=6.990 sample synthesized with 180. Also drawn is the susceptibility curve of the same sample after a site-selective isotope exchange with 160

140

E. KALDIS ~I0-~

I

0

-I

i i g

-2

-3

°~

-4

-5

o ÷

o +

-6

o ÷

2'o

o ÷

"~.'o

o

o

Y

o÷ o• ÷°+ +

go

i•

1~o

12o

TEMPI~RATOP.I~ [K]

Fig. 86. Morphology of the magnetization curve o f an equilibrium sample due to the splitting of the diamagnetic transition. The arrow shows the change of curvature. Open circles, YBa2Cu~806.990. Crosses, the same sample after site-selective isotope exchange with 160 which slightly increased the O content to 6.991. After data of Zech et al. (1995a) and Conder et al. (1994a).

2

X10"4

0 -2 -4 e~

-6 r,,,

r~

-8 -10 -12

87

;8

;0

;0

¢,-

~

93

;4

;5

TEMI~RATURE [K] Fig. 87. Magnification of the curves o f fig. 86 to show the extreme reproducibility of the equilibrium samples even after the series o f manipulations necessary for the isotope exchange. The arrow shows the temperature of the change o f curvature. After data o f Zech et al. (1995a).

and a slight increase o f oxygen content to x = 6.991. The change o f slope leading to To2 is clearly shown (arrow). Figure 87 shows a magnification o f the two curves o f fig. 83 in the higher temperature range. As can be seen, the reproducibility of the sample in spite of the manipulations for the isotope exchange is extremely high, the small change

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu3Ox 0°6

,

,,

141

!

•

x= 6.912

<>

x= 6.983

I

i

l (

0.4

.. o

0.2

O?O olo

*

oo/°! Ooox. 0.0

'~ i

O

i

m

i

I

a

.

i

..

s5.oo

9o.oo 95.00 T (K) Fig. 88. Derivatives of the magnetization curves showing the position of the Tc's.The optimally doped sample (x=6.912) does not show the splitting. The overdoped sample (x=6.983) does show the splitting with To2~ T~,opt ~ 91.8 K and Tcz= 89.5 K. After Conder et al. (1994a). being due to the actual isotope effect (Zech et al. 1995a,b). The flat part of the curve starts at ~88K. If a large number o f temperature measurements exists, then the first derivative of the curve can determine to high accuracy the two To values. Figure 88 shows the derivative peak (dMnorm/dT) of an optimally doped sample (x = 6.912) and an overdoped sample (x = 6.983). The latter clearly shows the two To's: To2 =91.8 K, almost overlapping with the T¢-optimal, and TeL = 89.5 K, the T~ decreasing with doping. As to the reason for this splitting we have seen in sect. 7.1 the ideas o f Loram et al. (1991). Another possibility is that the appearance o f two transition temperatures is due to the existence o f a pseudogap (T*) which appears mainly in the c-axis properties (resistivity, optical conductivity etc.) (see, e.g., Uchida 1997). We recall that the splitting o f the diamagnetic transition occurs at the displacive transformation at x = 6.95 (sect. 6.5), which is associated with structural changes along the c-axis (dimpling) (Conder et al. 1994a, R6hler et al. 1997a,b). The splitting o f the diamagnetic transition shows the existence of physical phase separation in the overdoped phase. Roughly, the stability o f the optimally doped phase is such that the material tends to retain it in cluster form even in the overdoped range. Thus the Tc2,onset stays constant near the value of the Tc, optimal. We will see that in the much wider Ca-123 "overdoped" range the effect is present to a much larger extent. 7.4. Phase separation due to stripe formation." separation of spin and charge As mentioned at the beginning o f sect. 7, a system o f doped holes in an antiferromagnetic insulator matrix tends to phase-separate. We have seen above the splitting o f the

142

E. KALDIS

diamagnetic transition in the O-overdoped phase, as evidence for the existence of phase separation. Before we review similar phenomena in the Ca-doped 123, we discuss briefly some experimental results concerning stripe formation in 123. One form of phase separation is the formation of stripes by the doped holes, separating hole-free, insulating, antiferromagnetic domains. The theory of the stripe-phase model is based on the theoretical work of Emery et al. (1990) and Zaanen and Gunnarsson (1989). First experimental proof of the existence of stripes in La cuprates and possibly other HTc cuprates came from the ingenious work of Tranquada et al. (1995) and, starting from another point, the work of Bianconi and Missori (1994) proposing the existence of quantum wires in a Wigner polaron crystal. The theoretical work of Emery et al. and Zaanen and Gunnarsson predicted the dynamic behavior of stripes and the existence o f cross-overs or phase transitions as a function o f temperature. As this behavior should be reflected in the lattice properties of the HTc cuprates, Sharma et al. (2000) used MeV helium ions channeling spectrometry to probe the existence of incoherent lattice fluctuations in 123-Ox at three different stoichiometries. This ultrafast real space probe has a response of <10 -17 seconds and can detect, in single crystals, static and dynamic uncorrelated atomic displacements with subpicometer resolution. Estimations of the spin-fluctuation frequencies in La-214 indicate strong temperature dependence in the range of 10 T3s at RT and 10-I° s at low temperatures, a time scale very well covered by the above method. For channeling to occur, the accelerated He ions are directed along a main crystallographic direction and undergo a sequence of glancing collisions with close-packed lattice elements (chains or planes). Many parameters influence the critical angle for channeling and are reflected on the measured full width at half maximum (FWHM) of the channeling angular scan. Most important - and very specific for the goal of this experiment - are incoherent displacements of the atoms from their regular lattice sites. Coherent displacements like those of thermal expansion are less important. Other important parameters are, e.g., the energy of the ions, the atomic numbers of the impinging and scattering ions, the interatomic spacing, and the electron screening potential. Sharma et al. (2000) used for their experiment single crystals with ATe < 1 K. Variation of the oxygen content was achieved with annealing in O2/N2 gas mixtures. Rutherford backscattering was measured between 300 and 30 K as a function of the tilt angle about the [001] axis of the crystals to determine the non-channeled amount of incident ions. The atomic displacements were extracted from the measured FWHM and were found to be appreciably larger than the Debye thermal vibration component. Subtraction of this thermal background led to the excess atomic displacements ue×. whose dependence on temperature with as parameter the oxygen nonstoichiometry are given in fig. 89. All three investigated stoichiometries with T~ = 45, 65 and 92.5 K show appreciable changes of slope at three different temperatures T1, T2 and T3, indicating cross-over or phase transitions, with values varying with the oxygen nonstoichiometry. For all investigated stoichiometries T3 ,~ Tc and the T/ and T2 are located in the normal state. The changes of the excess atomic displacements are most pronounced for the optimally doped samples (bottom), where the characteristic temperatures have the values Tl ~ 230 K, T2 ~ 150 K.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30 x

143

0.O6-

| 0.04'

0.02"

T3

•

.

=o

= e "

""

TI

~ T2

0.00

! 0.03.

! II

0.02.

"

T3

0.01

III

|

|

J

,t

TI

T2 0.00 0•03'

!

0.02

0.01

T3

•=

•

•

=" • • • _ '

0.00

go

0

T2

lao

T1

15o

2oo

2go

Temperature (K)

3oo

Fig. 89. Excess atomic displacements ue~ measured by ion channelling. Temperature dependence for three different stoichiometries, T~ = 45 K (top), 65 K and 92.5 K (bottom). The three characteristic temperatures support the "stripe phase model" (see text). After Sharma et al. (2000).

270j 210

ge 18°1 ...............

....

{

....

120

?

t

6O

°°t

30

6.5

6.6

6.7

6.8

6.9

7.0

x (oxygen doping concentration)

7.1

Fig. 90. Dependence of the characteristic temperatures of fig. 89 on the oxygen doping. T~, circles; T2, triangles; T3, squares; To, open diamonds. After Sharma et al. (2000).

Figure 90 shows the dependence of T1, T2 and T 3 o n the oxygen doping. A slight dependence is found for T1, nearly independence for Tz and the Tc dependence for T3. Sharma et al. (2000) discuss their results in conjunction with the stripe-phase scenario (Emery et al. 1997)• The assignment of T1 to the stripe formation temperature where

144

E. KALDIS

the holes segregate into stripes is not straightforward. The slight dependence on oxygen doping of T1 is in contrast with a more pronounced dependence predicted by this model or with the excitation energy gap from Giaever tunneling and photoemission measurements (Deutscher 1999). The authors point out, however, that the lattice fluctuations they study with channeling are charge fluctuations. On the other hand the tunneling and photoemission experiments detect fermions wifll both charge and spin. The difference between the channeling data and these experiments indicate that different energy scales are associated with the spin and charge quantum numbers in the normal state. One may, therefore, conclude that a non Fermi-liquid scenario with spin-charge separation is valid. The assignment of TI as the stripe formation temperature helps to understand the increase of Uex. with decreasing temperature (maximum between T1 and T2, fig. 89, bottom). Bianconi et al. (1996) have found that the formation of stripes leads to two different regions having different lattice constants. Two different lattice constants would enhance the excess displacements. With further decrease of temperature, Uex. decreases until it reaches T2. The lack of x dependence of T2 is in very good agreement with the temperature Jbr pairing within a single stripe predicted by Emery et al. (1997) or the pseudogap temperature found experimentally (Batlogg and Emery 1996) at nearly 150K. This would mean that an increase of Uex should be expected at lower temperatures when superconductivity sets on in the individual stripes. This is shown in fig. 89 (bottom). At Tc = T3 we expect an increased coherence of the lattice and this is manifested in an abrupt drop in ue×.. The increase at T < T3 can be assigned to spin-fluctuation dynamics, an indication of slowing boundary fluctuations which can be followed easier by the lattice. This work of Sharma et al. (2000) supports the spin-charge separation scenario as the superconducting properties of 123-Ox evolve with oxygen doping. Additional support for the stripe-phase scenario is coming from neutron scattering experiments by Mook et al. (2000), culminating a series of important papers by this group (Mook et al. 1993). Thus Dai et al. (1998) and Mook et al. (1998) were able to show the appearance of spin fluctuations, and Mook and Dogan (1999) that of charge fluctuations, supporting the existence of a dynamic stripe phase in 123-Ox crystals with high To. The interpretation of these results was, however, ambiguous due to the fact that the magnetic fluctuations due to the spins had a fourfold satellite pattern at incommensurate points around the (~, ~ ~) 1 magnetic reciprocal lattice position, which allowed also the possibility of nested Fermi surface. On the other hand, for a 1D stripe phase only one set of satellites is expected. Tranquada et al. (1995) and Tranquada (1998) proposed that the fourfold symmetry could be the result of stripes alternating in direction for consecutive planes along the c-axis. Mook et al. (2000) solved this problem after recognizing that the fourfold symmetry is the result of the twinning of the large crystal (25 g) they were using in the past. By applying pressure they could partially de-twin a smaller crystal and discriminate the single sets of satellites resuking from the two twin orientations. They showed for 123-O6.6 that the incommensurate magnetic scattering has a 1D modulation vector. Based on this result Mook et al. (2000) revisited their results on charge fluctuations obtained by investigation

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN Y B a z C u 3 0 ~

145

of the phonon spectra (Mook and Dogan 1999). They were able to show that these results also are consistent with iD charge fluctuations. The work of Sharma et al. (2000) and Mook et al. (2000) supports the dynamic stripe theory against the Ferflai-liquid approach (Zaanen 2000). For materials research it is important to know that stripe phases appear also in 123-Ox, in order to scout for possible mesoscopic structural indications of their presence as a function of nonstoichiometry.

8. O v e r d o p e d p h a s e - III: p h a s e s e p a r a t i o n in the C a - o v e r d o p e d r e g i m e ,

Y(Ca)-123-Ox Already at the beginning of the HTc era, Ca doping was assumed to increase the carrier concentration in 123-O~ (Tokura et al. 1988). Using the simple electroneutrality condition discussed in sect. 5.1, Q + 3(2 +p) = 2y,

(5)

they assumed - and after them a whole series of other authors (sect. 5.1, e.g., Tallon et al. 1995) - that the decrease of the positive charge due to the substitution of y3+ by Ca 2+ will result in a lineal" decrease of the oxygen content. In all these investigations, the decrease of Tc was considered as the only evidence for overdoping. All these studies of Ca-doped 123-Ox in the literature have been performed on samples that had not been fully oxygenated, and none was supported by a direct chemical analysis of the oxygen content with a high-resolution method. Btttger et al. (1996) presented the first systematic study on fully oxygenated samples with neutron diffraction. The samples have been synthesized by a solid-state reaction, similar to the CAR samples (sect. 3.1.2) to achieve maximal oxygen loading. Stoichiometric parts of highpurity powders (Y203, CaCO3, BaCO3, CuO) were mixed in ethanol and homogenized in a ball mill, then dried for one day (100°C). The mixture was subsequently annealed in air at temperatures 860°C (to avoid formation of eutectics), 870 and 885°C for 16 h. Further annealing in flowing oxygen at 900°C (8 h), 920°C (8 h), 950°C (70 h), and 960°C (30 h) was performed in order to expel as completely as possible the carbonate traces. To improve the oxygen diffusion and homogeneity, thorough grindings were performed between these annealings. To achieve maximum oxygen contents a very slow cooling rate (4°C/h) was used from 500°C to room temperature. A structure refinement (XRD and NPD) as a function of the Ca and oxygen content has been performed on 11 Ca-doped compositions (Ca 2-20%) reaching up to the solubility phase boundary (20%) and one highly O-overdoped sample (table 10). The decrease of Tc down to 55 K and two significant structural changes with increasing Ca-content, - a large increase of the apical bond by 1.25%, and - a decrease of the orthorhombic strain by 30%, supported the generally accepted picture of overdoping.

146

E. KALDIS

8.1. Oxygen content." the A and B states In the first exact investigation of the oxygen content (B6ttger et al. 1996) using highresolution volumetric determination (sect. 3.1.3), the linea( dependence of the oxygen content on Ca doping was proven not to be correct (Kaldis 1997). As mentioned earlier, this powerful analytical method has a very high accuracy, Ax = -4-0.001, due to calibration with the coulometric decomposition o f water; it allowed, therefore, for the first time an exact investigation o f the dependence o f the oxygen content on the Ca doping. Figure 91 shows this dependence which strongly deviates from the dotted line predicted by eq. (5) (Tokura et al. 1988) as shown recently (Kaldis 1997). This result shows that in spite of

the large Ca doping the oxygen content decreases only slightly and oscillates between two "states"A (x =6.980) and B (x = 6.966). Thus, starting with an oxygen-overdoped (O-O) 123 material having a very high oxygen content, x = 6.994 at (0% Ca), the oxygen content at 17% Ca reaches only x = 6 . 9 5 2 instead o f the expected 6.905 (the 20% Ca samples were not analyzed due to a small amount o f a second phase found by X-ray diffractometry). Almost all of the decrease of oxygen in the range 0-15% Ca is reached already at 2% Ca. It is also interesting that the phase boundaries o f the overdoped phases are approximately the same (x ~ 6.95 and 6.98) both for the Ca-overdoped and the O-overdoped samples, fig. 92. Application of oxygen pressure (100 bar at 900 K) on the oxygen-poor state B (at 10% Ca) (x = 6.96) shows that it can be further oxidized. This oxidation stops exactly at the A state (x = 6.98), indicating that this is the state with the highest possible oxygen content, which the B states tend to reach (fig. 93). For Ca% ~> 15 7,00

,i

•

6,99 ~

i

i'

i

' i i

".i

•

i

,

i

•

:1

• ,|

•

L

•

i ,---~, I ,.-

6,98 ~

"'~i

6,97

oii

.

X 6,96 "E

,

= 2"]

Q*3i2*p)

m~

b ,,

6,95 e" 0 6,94 U ¢,,- 6,93 ~ 0

I

/'".i i

0•6,92 6,91

,

""

i

i& Ca-t23 i

6,90

6,891 6,88

O~

i C,a-B I

I

-2

0

i

,i

2

. I ¢: Ca-BiCa-A

Ca-A I

4

•

i

,

6

i

8

Ca%

'

'

I

•

I

I

'

I

•

10 12 14 16 18 20

Fig. 91. Change of the O content x of Yl_yCayBa2Cu30~ (Ca-123-Ox) with Ca%. This is the first direct oxygen determination in the literature for this system. The assumption that the linear relation of eq. (5) is valid, is not correct (see text). The linear dependence used in the literature is also shown (diagonal dotted line). After Kaldis (1997) and Kaldis et al. (2000b).

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30 ~

Oxygen content Ca-123 ---7,02 7,00 6,98 6,96 6,94

7,04

7,06 3,8192

•

•

'3,81 0J

.

=" 3,8188]

::

3,81-1

;

~'L '~ E ~

3,8182-1 i 3,8180" ti 3,8178~ c 3,8176: 3,8174 2 3,8172 ,

~ i - ~

147

' '

~

i

~

3,8145 II~ 17%

If///.

3,8140"~. 3,8135

3,8130 V/,,2

3,8125

.0% •

'

i

i

•

,

•

'

,

• •

t

,

,

•

i

3811 g •

,

i

,

6,94 6,96 6,98 7,00 7,02 7,04 7,06 Oxygen content 123-0x

l Lay;rCa'123j

Fig. 92. Oxygen-poor phase boundaries of the overdoped phases 123-Ox and Ca-123-Ox in the example of the a-parameter: bold vertical axes at x ~ 6.95. Oxygen-rich phase boundaries at x .w.6.98. Lower x-axis: oxygen content of 123-O~. Upper x-axis: oxygen content of Ca-123-Ox. The oxygen has been directly determined with ±0.001 accuracy. After Kaldis et al. (2000b).

the oxygen content seems to follow the slope of eq. (5). The oxygen content of the Ca-free sample (oxygen-overdoped, O-O state), synthesized under the same conditions, is appreciably higher (6.994). We also note that in fig. 93 and table 10 a hysteresis appears for the A --+ B transitions at 9-10 and 14% Ca. In view of these unexpected results of the near independence of the oxygen content on the Ca doping, the question of the homogeneity of the samples must be discussed. It is known from the literature that it is very difficult to obtain homogeneous Ca-doped 123 samples, one of the problems being the substitution of Ca both for Y and Ba (Buckley et al. 1991). The substitution site depends on the chemical history of the samples. Thus, polycrystalline samples slowly cooled (B6ttger et al. 1996) show, up to 20 wt% Ca, no substitution o f the Ba site. This has been confirmed for the same samples by R6hler et al. (1999a) with Ca-EXAFS. On the other hand, single crystals, probably contaminated with some AI, do show, above l 1 wt% Ca, substitution of both the Y and Ba sites (B6ttger et al. 1997). Similar results have been achieved for single crystals grown in zirconia crucibles by Schlachter et al. (1999). This anomalous dissolution can be explained with the percolation limit of the Ca-123 clusters (sect. 8.3.1), which overlaps with 11%. Obviously for single crystals, a segregated phase with higher Y substitution is less energetically favourable than a combined Y and Ba substitution. Rietveld refinement with XRD is a very sensitive method for differentiating between Ca and Ba or Y, because the numbers of electrons are quite different: Z = 20 (Ca), 39 (Y) and 56 (Ba). Combined NPD and XRD also clearly show that slowly cooled, highly oxygenated, polycrystalline samples are single phase up to 17 wt% Ca. A small amount

148

E. KALDIS

7,00

'

~

• !

'

I

. . . .

I

'

"

~,

i Ca - 123Ox

'

::

"~-O i B6ttger et al.i 1996 i i l

6,99 ', \, iKaldis1997 X

"~ 6,98

"',

'\i

A

;

iA

tJ 6,97 ',

~

I/

,, i

6,96

6,95

"..

B

r

~:'d

,,.

B

Q+3(2+p) = 2~ \

s'

' ' 'to' Ca

' '

t~

' 21

% ~

Fig. 93. Oscillatory change of oxygen content vs. Ca%, as determined with high-resolution oxygen analysis. Application of O pressttre (100bar of 02 at 900K) on the oxygen-poor state B leads to its oxidation to the A state (arrow, 10% Ca). Note hysteresis phenomena near the A B transitions, with increasing Ca%. The dotted and dashed lines are guides to the eye. After B6ttger et al. (1996), Kaldis (1997) and Kaldis et al. (2000b).

of a second phase appears at 20%. The Ca content is unambiguously determined with these structural investigations (within 4%) and in agreement with the composition of the starting material. The incorporation of Ca is further demonstrated by the above-mentioned strong decrease of Tc and the expansion of the apical bond. We note, however, that the diffraction and magnetic measurements have large length scales. Microprobe analyses of cold-pressed tablets have been performed for the 9% and 14% Ca compositions where the A - B changes appear and the O analysis (table 10) shows some scattering. For comparison the 4% and 17% Ca samples were also investigated, together with the undoped (0%) sample. The results followed the calculated changes of the Y, Ba and Cu contents with increasing Ca doping for substitution of the Y by Ca (figs. 94-96). The spread in the measured values for a given Ca content shows the degree of homogeneity of the metal sublattice. The deviations are negative for Ba and Y contents and are limited to 3% or less. In fact, deviations of 1.5-3% appear only for the 9% and 14% Ca compositions near the transitions of the oxygen content A ~ B (fig. 93), the 4% and 17% compositions having much smaller deviations (0.6-0.7 and 0.1-0.5% respectively) supporting a hysteresis associated with these transitions. In addition, for the Cu content a small systematic negative deviation is found (fig. 96), possibly due to the well-known Cul defects. Some slight excess of Cu exists in the transition compositions

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBazCu30 ~

149

Table 10 Volumetric determination of oxygen (Conder et al. 1989) in Ca-doped 123 ~ Ca content (%)

Oxygen determination 1st Analysis

0

2nd Analysis

3rd Analysis

6.994±0.001

6.995

6.965

6.966

4

6.979

6.980

5

6.978

6.980

6.978

7

6.980

6.978

6.985

6.980

6.972

6.965

6.960

2

9

B -+ A

A--~B

10 12

B -~ A

6.963

6.963

14

A -+ B

6.972

6.978

15

6.959

6.960

17

6.952

6.955

20

-

-

" All samples were analyzed twice, except two analyzed three times. After data of B6ttger et al. (1996).

0 ;

14

13

........

! ...... •

~" "~

5 =

.:

0.6% 12

10 il

15 i

,

14

13

.... i-..I

......

o.~%

i .......

=

h •

:=

('.-i..~

Z

15"

"- "--.'

Theor. Y %

i--~=.

g=

i !

12

e

, • 1.2%

11'

....

.............. cl

i

i

11

o.2%

i

Ca-423

>.

20

spreading

10.

10 all 2.3%

9.

j 0

i 5 Ca % (Synthesis,

9 10

115 XRD,

20

ND)

Fig. 94. Microprobe analysis for Y as a function of Ca content. The theoretical Y content and the spreading of data for each Ca content are also given. Note increased spreading near the transitions. After Kaldis et al. (2000b).

(9, 14% Ca; A) which are compensated by slightly less Ba and Y. This nonstoichiometric disorder is small enough to evade detection by the structure refinements and the EXAFS investigations. Further discussion of the disorder of these samples due to ageing in sect. 8.3.1.

150

E. KALDIS

42,0

5

0 i

,

10

i

15

20 42,0

i i

41,5 -

Theor. Ba %

--

)

.......... . . . . . . . . . . .

hI

I

....¢.-I.........

i

. . . . . . . . . . . . . . . . . ..............

41,0 -

i

HI

"~2 40,5e~ e ' .2 40,0-

gi • 1%

d •

i

be

:. )

J

.

41,5

......... :

41,0

• O,~°~ spreading

40,5

i i

am

40,0

1.5% :. 39,5-

39,5

Ca-123

39,0 -

39,0 i

fn~ 3%i

38,5 -

38,5

i

0

5

10

20

15

(Synthesis, XRD,ND)

Ca %

Fig. 95. Same as fig. 94, but for Ba content see text). After Kaldis et al. (2000b). 0 i

5

10

i

i

'

31.

20 I31

'

fi~

Ca-123

/i

15

al

30 ¸

3O

gI!

P 0 29 ¸ Theor. Cu % . U ....................

. . . . . . . .

i ..........................

.......... "

I 0.1%

hm)

28

i

I

i

1% i

0

d•

I i

28

spreading

el

3°/~

i 27

i ..........

29

3% 27

0.6% i

5

II0

1=5

20

Ca % ( S y n t h e s i s , X R D , ND)

Fig. 96. Same as figs• 94 and 95, but for Cu content (see text). After Kaldis et al. (2000b).

The instabilities leading to the changes between the states A and B (figs. 91 and 93) were considered to be a kind of phase transitions (Kaldis 1997). The compositions at which they appear are marked in all figures by vertical dotted lines. Figures 91 and 93 do not show the average of the analytical oxygen determinations of each sample, as usual in the literature, but the results of the individual determinations for each batch (see also table 10). Owing to the high accuracy of the analytical method deviations above the analytical error Ax=~0.001 tend to indicate statistical deviations in the homogeneity of the batches. Whereas very good agreement is achieved between the two oxygen

OXYGEN NONSTOICHIOMETRYAND LATTICEEFFECTS IN YBazCu3O,

151

determinations for most samples, indicating good oxygen homogeneity, clear deviations are found near 9-10% Ca and at 14% Ca, in agreement with the nonstoichiometric disorder found by the microprobe investigations. This spreading of the data can be attributed to hysteresis associated with the A ~-~B transitions. Additional support for the existence of transitions or cross-overs associated with the change of O content is the fact that miscibility gaps could not be traced up to now with XRD and ND (Brttger et al. 1996). It seems, therefore, probable that B --+ A transitions take place in the ranges 2 - 4 % and 12-14% Ca and A -+ B transitions in the ranges 9-10% and 14-15% Ca. The solid or dashed lines and arrows (guides to the eye)joining the data in figs. 91 and 93 illustrate this possibility. The question arises now, what is the reason for the unexpected deviation of the oxygen content from the predictions of the electroneutrality condition? Inspection of eq. (5) shows that changes of the oxygen content may arise for two reasons: either changes of the charge of the cations or of the charge per (Cu-O) +p pair. Taking into consideration the anomalous incorporation of Ca into the lattice, discussed in sects. 8.3 and 8.4, it seems possible that such changes appear due to screening effects" of the charge of Ca2+. Variations of this screening with increasing Ca content may lead to the A+-+B changes of the oxygen content. The spreading appearing in the oxygen and heavy-metal contents (figs. 94-96) at these changes, can be considered to result from the hysteresis of phase transitions triggered by the variation of the effective Ca 2+ charge. These transitions and the general deviation of the experimental curve from eq. (5) would then be associated with changes of the localization of holes with Ca doping (sects. 8.3, 8.4) which could proceed stepwise. Thus, the A ~ B transitions may be attributed to slightly different oxygen ordering configurations triggered by the Ca distribution. It is due to the extreme accuracy of the oxygen determination that they can be detected. 8.2. Structural optical and magnetic characterization of Ca-123-Ox In view of the extreme resolution of the oxygen determination, Ax -- !0.001, due to the coulometric calibration (Conder et al. 1989), it is clear that many physical measurements do not have the necessary resolution to trace effects of that order. In characterizing the properties of the Ca-doped samples, the only ones "fully" oxygenated in the literature, it is reasonable to look also for indications of the existence of the A and B oxygen content states.

8.2.1. Structural characterization Plotting structural parameters and some bond lengths vs. oxygen content leads to segregation of the data into two groups corresponding to the A and B states of figs. 91 and 93. Figures 97a-c show the dependence of the a-, b- and c-parameters (Brttger et al. 1996) on the oxygen content (Kaldis et al. 2000b), and figs. 98a-c show their dependence on the Ca content (Brttger et al. 1996). In fig. 97 there is not much scatter in the data of each group, almost none for the c-axis. In fact, the linear scaling of the lattice parameters with the two groups of oxygen

152

E. KALDIS '

i

,

i

.

i

,

e

3,8145-

J

3,84

,

i

(a)

10°, ~ 3,8140- o_=17~° ',

A 7% Io

E 3,8135 12°,'~ =

i i

i

'

3,821

'

3,8130- 15°,~! ~2% i

l

g

i

~ 3,61-

+0.085

-

%

i

9%J

~ 3,8125-

Ca-123 BOffgeret al. 1996

E 3,83-

o--~14O/o

L_

E

f

(a)

i i

'~' 3,8120-

Ca-423

3,8115-

d,= 5%

[ •

a.pararn 0-2 ]

I o

a.param 0-1 I

•

i

.

i

,

3,80

O-C

3,890

i

,

i

Oxygen content ,

.

,

.

9"2%

i

(b) -0.23 %

,

E

O-O

io, 4%

E .~ 0

,

A

3,880.

3,8703,865Ca-123 B0ttger et aL 1996

~i 3,860-

.0 3,855~

,,

3,878.

3,850

1'oCa % 1'~

i

E t

11.680

o12% t "15%

~,. 3,676.

'

'

t

.

.

.

.

i

....

;i

.

.

•

i

.

.

.

i

,

~11,67o! Ca-123 B0ttger

Ca-123

f

.

(c)

11,675-

17%[

3,874 •

2'0

14%

i

.6

2'o

E 3,860:

x

(b) o.0,

B

1'5

3,885 ~

3,875 ~ .

Ca %

,

6,95 6,96 6,97 6,98 6,99 7,00

3,882

lb

6

0O/o~i

~11,666- et af.

1996

6,95

6,96

6,97

6,98

6,99

11,660-

7,00

Oxygen content

i•

+1.75

E 11,656-

x

~..11,650i

O i

11,675-

,

i

,

i

Ca-123Ox

(c)

11,640

17% 11,670',,B

~

11,665 !

~

t 1,860-

.~

11,655-

6

15%'Wz

o

.

10%

A

". 7% 9% '~;~-n ~5%

12% ".p

4% 11,650 11,645 6,95

0-0

2°/oi~ '

i

6,96

,

0%~ I

6,97

,

i

6,98

A

B 0

5

~ B 10

A

Ca %

15

20

Fig. 98. Lattice parameters of Ca-123-O~ vs. Ca%: (a) a; (b) b; (c) c. Note that a is almost independent of Ca%, b changes little, and only c increases appreciably (c£ table 6; see text). After data of B6ttger et al. (1996).

14%

n.

11,645-

,

i

6,99

Oxygen Content

x

7,00

Fig. 97. (a) a-parameter of Ca-123-Ox as a function of oxygen content x, for various Ca contents (given in %). (b) b-parameters. (c) c-parameters. Note the good scaling with oxygen content. The data segregate into two groups corresponding to A and B. Circles (squares): 1st (2nd) O determination. After Kaldis et al. (2000b).

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa~Cu30~

153

Table 11 Relative changes of the lattice parameters and apical bond with doping Doping

Relative change (%)

123-O~ Ca-1230~

a-parameter

b-parameter

c-parameter

Apical bond

-1.15 +0.085

+0.5 -0.23

-1.45 +1.75

-7.6 +2.8

2,300

B

g o, C"~

2,295

A

B AiBI

iCa-1230, i eBtattl!er ! i1996

2,290

2,285

/ I

~"

O "1~

e,, o

2,280

2,275

m o

mm

2,270-

2,265

J

' I '' 0

'"

i .... 5

'l . . . . . . . . 10 15

Ca %

= ' 20

Fig. 99. Change of the apical bond with Ca%. Slope changes at the A - B transitions (vertical dotted lines) can be seen at 4, 9, 12, 14 and 17% Ca. After data of B6ttger et al. (1996).

content (A and B) is remarkable. Figure 98 shows the expected opposite trends to the O doping shown in fig. 26, supporting the current ideas of overdoping. It is interesting to note that the relative changes of the lattice parameters with doping are appreciably different for Ca doping (0-20%) and O doping (x = 6.5-7.0) as shown in table 11. We may, therefore, conclude that Ca doping, in contrast to O doping, affects only slightly the a-axis, to some degree the b-axis, and more strongly the c-axis. This shows rather small interaction with the superconducting planes and the oxygens in general. The expansion of the c-axis is also due to the larger Ca ion. The apical bond is also less affected by the Ca doping. An influence of the A and B states may be seen in the plot of the apical bond vs. Ca% (fig. 99), a bond length crucial for the charge transfer (sect. 3.2.1). Five different ranges appear which seem to be related to the B and A states (vertical dotted lines). The increase of the apical bond is interrupted in the range 10-14% Ca and continues with higher slope at >14% Ca. As we will see, changes in the 14-15% range appear in most properties (magnetic, Raman, EXAFS).

154

E. KALDIS

2,30 o~

17% ~

~

2,29-

Ca-123

B

i

A

l

0•

2,28-

\c= 15% \ 12%=, =-o \ 10%

',14% C'~

o 7%

,, ,,-e 9%

i

e~ 2,27-

----

= ',

5% b, 4%

i

(!

i i

.m

O.

c='\ 2% 2,26" '

6,95

I

6,96

'

I

6,97

•

6,98

I

6,99

Oxygen content x

7,00

Fig. t00. Scaling of the apical bonds with oxygen content for various Ca%. The data segregate very well into the B and A groups. Structural data from NPD at 1.5 K (B6ttger et al. 1996). After Kaldis et al. (2000b).

Similar to the lattice parameters also the apical bond scales very well linearly with oxygen (fig. 100). 8.2.2. Raman characterization The advantages o f micro-Raman measurements for such investigations were discussed in sect. 5.5. We mention here that not only macroscopic but partly mesoscopic statistics of the distributions of phases as a function of the dopant become available. Recent micro-Raman investigations o f these samples (Palles 2000, Palles et al. 2000b) showed the existence o f phase separation with Ca doping. Several new modes were observed, some scaling with Ca content and some scaling with oxygen content, indicating phase separation. In addition to the well-known Big mode (out-of-phase vibrations o f the plane oxygens 0 2 and 0 3 ) at 340 cm 1, another mode with the same symmetry appears at ~322 cm-~; its intensity increases with Ca content. More interesting in the present context are the changes appearing in the Ag phonon (in-phase vibrations o f 02, 03). As we have discussed earlier, this phonon is associated with the dimpling of the superconducting planes which undergoes appreciable changes in the oxygen-overdoped range o f 123-Ox (sect. 6). In addition to the main in-phase mode at 429 cm -I a new mode appears at - 4 0 2 c m 1; its intensity increases also almost linearly with Ca content. But it is the main in-phase mode which seems to reflect the changes between the A and B states:

O X Y G E N N O N S T O I C H I O M E T R Y A N D L A T T I C E EFFECTS IN YBa2Cu30 x

440

'

I

. . . .

I

. . . .

I

'

'

155

'

|

E

2~2

.~0435-

?222

~430. -,1 O" ~ 45 ! e-

~

NNN

40

Y~_yCayBa2Cu30 x Palles et al. 2000

e-

"~

35

tO

I

e- 30 O Q. • 25

t-

,,C

~e- 20

?)72 '

I

0

. . . .

I

5

. . . .

I

'

•

•

10

Ca content y

'

15

[%1

'

'

I

20

'

Fig. 101. Raman in-phase phonon ( 0 2 , 0 3 vibrations) width vs. Ca%. Note the abrupt increase of width above 15% Ca. After Palles (2000) and Palles et al. (2000b).

- Its width increases appreciably above ~14-15% Ca due to an increasing contribution of the 402 cm-I mode, indicating the possible existence of a phase transition coinciding with the last A---+B transition and the end of the oxygen content oscillations (figs. 91, 93). Figure 101 shows this effect. - Particularly at low temperatures a splitting appears which could be fitted with three Lorentzians at 402, 429 and 439 cm 1. The relative intensity of the main in-phase phonon peak 429 cm -1 to that of the 429+439 peaks seems to scale with the oxygencontent variations between the states A and B but not with the Ca content. Figure 102 shows this dependence of the relative amplitude on Ca%. Five to seven crystallites have been measured from each Ca composition. Clearly the relative intensity does not scale with the Ca content, but it seems to scale fairly well with the oxygen content (figs. 91, 93). The frame of the dotted and solid lines and the arrows shows an idealized behavior which is compared with the data, and it is only a guide to the eye. Again, the spreading of the values for different crystallites of the same Ca content at 10, 12 and 15% Ca exists, which can be attributed to hysteresis and pinning effects due to the existence of transitions between the A and B states (see sect. 8.1). Additional evidence exists for the transition at 14-15% Ca (the last A - B transition) from other measurements:

156

E. KALDIS

YvxCaxBa2Cu3Oz_y I

B

~7,00 O

...............

6,96

I

i .......

I

A

B

l

:

'

I

iA

............

..............................

-

A

Micro-Ramani Palles et al. 2000 ~

0,8. ......................

,~

,

..........

0,6,

e" 0,4 i i

__e-

I

o

I

I

1; Ca c o n t e n t x [%] ~

~

~'~

2'o

Fig. 102. Upper part: oxygen content vs. Ca%. Lower part: comparison of the relative intensity of the Raman in-phase modes 429/(429+439) vs. Ca% with the A B transitions of the oxygen content. The 439 cm i mode is a new mode appearing with the Ca doping; however, it does not scale with Ca and it seems to scale with O. After Palles (2000) and Palles et al. (2000b).

- The apex phonon width shows an appreciable increase at 15% Ca due to the appearance o f new modes. - The To vs Ca% shows an anomaly in this range (see below). - The dimpling (Cu2-O2,O3) vs Ca% shows a transition both in neutron diffraction (B6ttger et al.) and EXAFS (R6hler et al. 1999b) measurements (see below). The latter changes slope at 9% and 15% Ca. 8.2.3. Magnetic characterization As discussed earlier (sect. 7.1), oxygen-overdoped 123 shows two critical temperatures (see e.g., fig. 85). The same effect is observed in Ca-123 (Kaldis et al. 2000b).

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30~ 10%

0.2

i

"

G2Z7 a

- •

I

Polynomial •

i

~4:33

0.0

Fit of G2Z7 =[

71.4

Po~tlOm~RegressionforG2Z7norm:

LO

v

-0.4 -0.6

Iv

F~rameter

P~

Error

Value

A

-1.01289 0.02101

BI

0.00334 0,00499

B4 B5

-2.39841E-7 2.02839E-g

~,~

;~?,;~ ~:;,:,'~'.,-'

,~

norm

|

•

~"~

v = A + e , x + a~'x'~ + ~.x,s + ~'x^~ * ~'x"5

-0.2

/

K

157

88.0

K

--

/

1.8468E-7 9.g4759E-I 0

/

.

/

/

-0.8 -1.0

-1.2

I

0

.

I

,

20

I

40

i

I

i

60

I

80

"

100

Temperature (K) Fig. 103. Normalized magnetization curve for Ca-123-Ox with 10% Ca. The onset Tonset=T~2 is 88.0K. Polynomial fitting results in Tel =71.4K. After Rossel et al. (1999) and Kaldis et al. (2000b).

Figure 103 shows a typical normalized magnetization curve for 10% Ca. The change of curvature at the higher temperature range leading to a second To (Tcl) is clearly seen (cf. figs. 86, 87). Due to the wider transitions of the Ca-doped samples, To2 was not determined by the Shoenberg method (Zech et al. 1994, cf. sect. 7.3) but by a polynomial regression (Rossel et al. 1999). In this particular case, the onset temperature was found to be Tc,onset~Tc2=88.0K, and the fitting gave Tcl=71.4K. Figure 104 shows the dependence of both these temperatures on the Ca content. Whereas Tel changes drastically down to 47 K, To2 remains near the Tc value of the oxygen optimally doped and overdoped 123 region, changing very little even for large Ca concentrations. As discussed in sect. 7.3 the appearance of two Tc's is evidence for phase separation. It can be argued that similarly to the O-overdoped region, the optimally doped phase is thermodynamically more stable and it continues to exist even under higher overdoping leading to physical phase separation in optimally doped and overdoped clusters. It is interesting to compare the To of fig. 105 with the corresponding one for oxygen-overdoped 123 (fig. 85). The onset temperature To2 remains also constant in the range x > 6.95 and has the same values for both the oxygen-overdoped and the Ca-doped ranges. The splitting of the diamagnetic transition in the Ca-doped phase, similar to the O-overdoped phase, is a strong indication for physical phase separation. We note that if chemical phase separation were present, then the Tc's should form plateaus like those of fig. 45.

158

E. KALDIS lOO

•

,

959o i

-e

,

,

o

i

,

,

,

t

,

o ........ ~ T==

-5~

~5 .....

• i

.

!

o

. . . . . .

.........

85 i

i

80:

:

~'

• e

75

i_,~ 70 65 i

Ca-123-0~

60 i

\\\ib,N ~\\\3

55 i

50 i

B

A

•

~\'~\%]

l

" 2

4

6

8

10

'12 14

16

'18 20

22

6,99

7,00

Ca % 1 006,95

6,96

6,98

6,97

gf

I

A Tc2, °nset Di

90

Fig. 104. To, and TCz as a function of Ca%. Tc2,o,~t is almost independent of the Ca doping, behaving similarly to oxygen-overdoped 123-Ox and having approximately the same value (cf. fig. 85). Tc2 decreases strongly, and shows a strong anomaly at ~13-15% Ca (hatched area). At this concentration the width of both in-phase and outof phase phonons increases and the dimpling change of slope takes place (EXAFS). The strong splitting of the diamagnetic transition of Ca- 123-Ox indicates the existence of physical phase separation in the Ca-doped phase, similar to the O-overdoped phase. After Kaldis et al. (2000b).

: []

O-O 0

!

D

i - ~ - ~ - o ::

~ - -~- ~ 3

•

I I I

80 I I

70

t~ I

I--

I

Tc~ l

it

~

50

'

6,95

!/ i II

I I

!m

Ca-123

t/ t

K a l d i s et al.

it

I

60

Imi

i iii

I

0

ml ! II !

i i

i

i

6,96

•

i

6,97

2000b []

i,

mc2onset I

i

6,98

6,99

7,00

Oxygen content x (average)

Fig. 105. T~I and To2 dependence on oxygen content of the 11 Ca-doped samples of fig. 104. The data scale linearly in A and B groups, showing a correlation of the A and B states with Tc. After Kaldis et al. (2000b).

The Tct curve (fig. 104) shows clear anomalies with a discontinuity in the range 13-15% Ca (hatched area) which coincides with the B ~ A transition at 13-14% and A ---+B at 14--15% (cf. fig. 93), the abrupt increase of the in-phase phonon width (fig. 101) and of the apex phonon width due to the appearance of new modes. This is additional support for the phase-separation scenario. Figure 105 shows the dependence of Tc on oxygen content. Both Tcl and Tc2 scale with oxygen and the data segregate in the two states A and B, indicating a correlation to superconductivity. Figure 106 shows that the relative intensity of the Raman in-phase mode ratio 429/(429+439) (cf. fig. 102) scales well with Tc2, indicating that the splitting of the

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu30x

2 ,,,

85-

,,..,~,

159

A

8075~"

12",

'

% Ca

70-

I-'-° 6560-

F-

i

17 ' , 1 4 \

=

'

55- Ca-123 ~-~-~ ", .2d~/oca Pallesatal. ' - \ 502000

45J

0,4

0,5

0,6

0,7

0,8

0,9

intensity ratio 429cm11(429+439)

Fig. 106. Dotted lines: T~ scaling with the Raman intensity ratio 429/(429+439), of the samples belonging to groups A and B (cf. fig. 102) (see text). After Palles (2000).The solid curve connecting the data is discussed in sect. 8.3.1.

in-phase phonon is related to superconductivity. Also, this figure which is not directly related to the oxygen content, shows a good scaling of Tc with the data corresponding to the A group (4, 5, 7, 9% Ca) and a fairly good one with those corresponding to the B group. The correlation of fig. 106 is the only one where the 14% sample is grouped with the B group instead of A, indicating that the investigated crystallites have already performed the transition to B. This is an important fact, which supports the reactivity of the compositions at the A - B transitions. Up to now the decrease of T~ due to doping with Ca was considered in the literature to be due to the increase of the hole carriers. A possible localization of these carriers, as we will discuss in sect. 8.4, was not taken into consideration. 8.3. E X A F S investigations To investigate the controversial mechanism of Ca incorporation in 123, Y-EXAFS investigations have been performed in the same samples (R6hler et al. 1999a,b). Figure 107 shows the dependence of dimpling as a function of Ca content resulting from the EXAFS and the previous NPD (B6ttger et al. 1996) investigations. A wide steplike decrease of dimpling appears at 9% Ca and ends at 14% Ca (both compositions of A - B transitions), which could not be so clearly resolved with neutron diffraction. However, the total change of dimpling is the same for both methods. Thus, these measurements also support the existence of the A and B states, but put a questionmark to the number of transitions. This will be discussed in sect. 8.3.1. The EXAFS results concerning the incorporation of Ca were completely unexpected. Up to Refr ~ 5 A the EXAFS spectra of 123 show no substantial difference with or without

160

E. KALDIS 0,290

~"

0,2850,280:

o

0

.,

....

,...

: 0a-123 "

•

, ....

•

• _

-,

~

0,2,5-o.. o o

0,2zo i

i\

\o/\o

, ....

0,290 ~

ii

~u,z~u

!

t0,275

0 285 //i::R0hler -:J , iietal. 199gl ....

•

0,265-

0,270 0265

0,2604 ~. 0,255i

B0ttger et al. 1996

},~ 0 , 2 5 0 : 0,245

,..

i i ExAFS'

.,...., 0

A .... 5

°~i\~o -~ii ~ ~ =°::i~\

~ 0,260 -] 0,255 J

~il E o -~0,250 ~°i in, B :l , .... , .... ,. 0,245 10 15 20

Ca%

Fig. 107. Y-EXAFS investigation of the dimpling (Cu2-O2,O3) vs, Ca%. The decrease of dimpling changes slope at 9% and 14% Ca. After R6hler et al. (1999b). The transition at 14% is clearly supported by the Raman data (cf. fig. 101) and the Tc data (cf. fig. 104). For comparison the neutron diffraction data of B6ttger et al. (1996) are also shown.

Ca doping. The question arises, therefore, where does the Ca enter into the Y-123-Ox lattice and in what electronic state? - As mentioned already, both neutron diffraction (B6ttger et al. 1996) and EXAFS (R6hler et al. 1999a) have shown clearly - in contrast to the findings of Buckley et al. (1991) - that in equilibrium, polycrystalline, fully oxygenated samples Ca substitutes only for Y. For single crystals, which have a completely different chemical and thermal history, Ca substitutes for Y and Ba at higher concentrations (>11% Ca) (B6ttger et al. 1996, Schlachter et al. 1999). - In view of the strong decrease of Tc] with Ca doping and the neutron-diffraction evidenced dissolution in the lattice (e.g., c-lattice parameter change) it has to be concluded that Ca is dissolved in the 123-Ox lattice but in monodisperse conformation, i.e. with only Y as nearest and next-nearest neighbors (R6hler et al. 1999b). This could be possible in principle up to 25% Ca, but strongly disordered Ca ions segregate in clusters of up to 4 NN. Unfortunately, model calculations show that these clusters can be resolved by Ca-EXAFS but only for measurements with photoelectron momentum k >/12A 1 (R6hler et al. 1999a). This work is presently planned. The simplest straightforward model proposed by R6hler et al. is Ca-substituted 123 cells surrounded by the NN Y-substituted cells forming cross-like clusters of 5 distorted cells. Figure 108 illustrates this model for increasing Ca content. R6hler et al. (1999b) pointed out that if the single Ca ions percolate in a random way, the critical value for a square two-dimensional lattice would be in accordance with existing models (Stauffer and Aharony 1995), ~60% Ca. For the cross-like 5-cell clusters the critical percolation value would be five times smaller, i.e. ~12%. This is in fairly good agreement with the dimpling changes in the 9-14% Ca (fig. 107) range and also with the increases of the width of the Raman in-phase and out-of-phase phonons at ~15% Ca, found in the same samples by Palles (2000) and Palles et al. (2000b) (fig. 101). It is also in agreement

O X Y G E N N O N S T O I C H I O M E T R Y A N D L A T T I C E EFFECTS IN YBa2Cu30 ~

[ ~ . l ~ . ~ i . . ~ . ~.. ~ . -.:E~l.~,l:~I~ • F"~'-i,"/,/-'X~'X~J ," • ~ ~,i,~,

161

• "K~. • .li.V,.~.~-k,l~.l~.il.l.l.l.l.l,i,l*l.l.l ~ - , X , ~ . F,,'~H/~_,~.r~X,~A. ~ , x - ~ . i i . F ~ - ~ - I ~ ~- ~ r , ~ , , V ~ , , ~ I • I • I • I • I • I "1 • ] • I • I • II • I ' 1 • J• i • I • I • i • I • I ' 1

VcAv~F',~//~A.~A ~'¾4",I"4,',11• L3¢X,~d¢,,i• ~

• i~l • W-X ' 4"A. V~V4A. O'Av,.~• II*I~ • .t~CA• V,¢,I• V4,,1• V,-'/]

l~a~d~(-P~4X~'~l. *l~l" • "l~" " "ll'r/4~l'r-~#A'~d~l'll'l'l'l'l'l'l'l'l'l'l ~ • F~Z21 • ~i" V ~ " ~',em • r~l. t .I. t .t- H. t.t- t • a .1~ • W .1~. ~ • ~ ~. •~ • ._J ~ r ~ / , ~ ~ / - ~ ' ~ r ~ i • v~ •~ •~ • ii • I ' 1 • I ' 1 • I ' l • I ' 1 • I ' 1 I~vA/~,i~,L79Y,,F/7#~,*#/ll • - I ~ .

4%

• .1~.-

9%

-II.I.l.l.l.l.l.l.l.l-II.l,l.l.l.l.l.l.l.l.I

16%

25%

Fig. 108. Model for the monodisperse dissolution o f Ca-123 in an Y-123 matrix so that all N N are Y-123. Clusters in the form o f crosses, consisting o f 4 distorted molecules o f Y-123 surrounding 1 molecule o f Ca-123 are proposed. A percolation is expected at - 1 2 % Ca. This is in good a g r e e m e n t with the experiment (fig. 107). Based on this m o d e l one would expect that the holes introduced by the Ca 2+ into the y3+ matrix will be screened by the Y-123 NN. After R r h l e r et al. (1999b).

with the oxygen-content results (fig. 93) which show A - B transitions at 9% and 14% Ca and an end of the transitions at 15% Ca followed by a stronger decrease of oxygen. Last but not least, the dimpling change at 14% Ca is in excellent agreement with the step in Tc2 at 14% Ca (fig. 104) showing the influence of Ca percolation on the decrease of T~. More information on the mechanism of the incorporation of Ca in 123 could be expected if Ca-EXAFS investigations at k ~> 12 A 1 are performed. 8.3.1. The A - B transitions and ageing effects We may now draw the balance of the correlations of the various properties of the material with the A - B transitions: (a) Large-length-scale methods, XRD, NPD: The lattice parameters a and b change very little with Ca doping. The main change is in the c-parameter which hardly reflects the A - B transitions. The transitions become more visible in the apical bond (fig. 99) where changes of slope take place at 3-4, 9-10, 12-14, and 17-20% Ca. Plotted vs. oxygen, most structural data scale with groups A and B. The magnetization measurements T~] show clearly anomalies in the range of the transitions at 12-14 and 14-15% Ca. Deviations seem also to exist in the range of the 9-10% Ca transition. (b) Small-length-scale methods: Raman investigations give appreciable evidence for the existence of A - B transitions. At 15% Ca, splitting of the phonons characteristic for the vibrations (11 c-axis) of the plane oxygens 0 2 , 0 3 (dimpling), out-of-phase 0 2 , 0 3 (Blg) and in-phase (Ag) (fig. 101), takes place. The latter (439 cm -1) splits into 429 and 439 and their relative intensity scales fairly well with the oxygen content (fig. 102). On the other hand, as we have seen above, EXAFS show transitions only at 9% and 14% Ca. Small deviations of the data at 2%, 7% and 17% Ca are too small to be considered, although some appear also in NPD. A full reconciliation of all data would be achieved if the oxygen content of 2% Ca (B state) would be shifted to that of 4% (A state), and the 14% (A) in the opposite direction (figs. 91, 93). An error in the very well reproduced oxygen content (table 10) does not seem possible, as hundreds of measurements have shown up to now, with this extremely accurate analytical method (sect. 3.1.3). In view of the fact that each analysis uses 100mg of material,

162

E. KALDIS

the above results represent the statistical average of 200 mg which is representing the average bulk properties of the sample. More suitable for an explanation seems the ageing of the samples. The oxygen analysis, XRD and magnetization measurements (without the 14% sample) have been performed nearly at the same time (B6ttger 1994). The NPD which is the origin of the structural data presented here, was performed in 1995 (B6ttger et al. 1996), and did not show any difference from the XRD data. The Raman measurements were performed in 1997, the microprobe measurements and the EXAFS in 1998. Additional magnetic measurements have been performed later (Rossel et al. 1999) to check the quality of the samples (these gave Tc changes not larger than 1 K), and to determine the Tc of the 14% sample. It seems, therefore, reasonable to assume that during this time some changes in the oxygen composition and the disorder of the samples in the transition ranges could take place. Thus, whereas the analysis, the structural properties and the magnetization (without 14%) give the original picture of the properties, subsequent changes of the samples are mirrored in the other measurements. Particularly vulnerable seem the transition compositions (figs. 91, 93) because, as we have seen in the example of the 10% sample, they can be further oxidized. This would explain why no substantial change has been found with EXAFS for the 2% composition. The case of the 14% sample is more obvious. The microprobe analysis (figs. 94-96) shows that, at least after four years of ageing, this sample and that with 9%, in both of which EXAFS finds a change of slope, have a certain cationic disorder: an Y and Ba deficiency, compensated in some parts of the samples by more copper. The origin of this disorder is probably intrinsic, as already the difference in the two oxygen determinations (table 10) of these samples indicates. We have attributed this disorder to hysteresis in the A - B transitions involved. Therefore, it is reasonable to assume that given enough time this transition will proceed through the bulk of the sample. Then, with the 2% sample developing an x ~ 6.98 and the 14% sample an x ~ 6.96 the present picture of the samples, as given e.g., by fig. 106, would correspond to the solid zig-zag line with one A - B transition. Similar curves can be drawn also for other curves plotted vs. oxygen. The assumption of an ageing process is also supported by the appreciable scattering of the Raman data in the samples of 9% and 15% Ca composition. Summing up: EXAFS and NPD show that a decrease of the dimpling takes place between A and B, starting at 9% Ca and ending at 14%. This is the percolation process of the monodispersely dissolved Ca which ends at Ca ~> 14%. NPD and XRD show that the contraction of the apical bond ceases during this percolation. This shows consistency between older and recent measurements, and gives an explanation for two A - B transitions. The other transitions appear in the oxygen determination and the linear scaling of the NPD and XRD data vs. oxygen content. Their appearance is possibly the result of Ca ordering in the monodisperse solution. This process influences the mobility of the holes introduced by Ca. If they are trapped (R6hler et al. 1999b, Merz et al. 1998, Niicker 1999), as will be discussed in the next section, then it is possible that the oxygen content remains constant. Depending on the screening of Ca charge resulting from the conformation of its dissolution in the lattice (see above), variations of this charge may result, leading to changes of oxygen and the A - B transitions (cf. sect. 8.1).

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTS1N YBa2Cu30X

163

8.4. Ca doping and localized carriers From the beginning of the HTo era it has been assumed that Ca increases the carrier concentration, their negative interaction leading to a decrease of To (e.g., Tokura et al. 1988, Tallon et al. 1995, Bernhard and Tallon 1996). The picture emanating from the above-mentioned work of R6hler et al. (1999a,b) could be quite different: From the cloud of Y cells surrounding the Ca cells of fig. 108 it can be assumed that the holes introduced by Ca 2+ doping are trapped and remain localized. Therefore, not additional carriers introduced by the Ca are responsible for the decrease of the Tc, but the decrease of the superconducting carriers due to decreasing Y-1230x as the substitution with Ca increases. In case of no dissolution of Ca-123 in the Y-123 matrix (chemical phase separation) the two Tc's would be observed to form plateaus (cf. fig. 45). The exact oxygen determination of the Ca-123 samples (the first ever made) which led to the finding that A and B states have less effective charge of Ca than had always been assumed, gives a very important support to the above picture of R6hler et al. (1999b): different screenings of the Ca charge at compositions lower or higher than the percolation limit (monodispersely dissolved Ca in the 123 lattice). Islands of Ca-123 in the Y-123 matrix (fig. 108) are in general agreement with physical phase separation shown by the magnetic and Raman measurements, the similar behavior of the O-overdoped 123-Ox phase (sect. 7), the general scenario of the two-component picture of Gor'kov and Sokol (1987) discussed later by Mihailovic and Mtiller (1997), and the literature of phase separation. The stability of the optimally doped 123-Ox phase leads to a segregation of an optimally doped superconducting matrix (Y-123) with itinerant carriers and non-superconducting clusters with localized carriers (Ca-123).These results may change our understanding of the mechanism of Ca overdoping, and probably also of the validity for 123 of the "universal" Tc vs. doping parabola (Tallon et al. 1995, Bernhard and Tallon 1996) so frequently cited. Another very interesting investigation of Ca-123 supporting the localization of carriers and the two-component picture (Mihailovic and Mfiller 1997) was performed by Merz et al. (1998). They used polarization-dependent near-edge X-ray absorption fine structure (NEXAFS) measurements to investigate the unoccupied electronic structure of Ca-123-Ox samples with various degrees of Ca and oxygen doping. The investigations were performed on detwinned single crystals. Their results have shown that, as expected from the immediate vicinity of Y to the CuO2 planes, the holes produced by the substitution with Ca appear only in the planes. This is in agreement with optical measurements by Widder et al. (1996). Independently of the Ca or oxygen doping the holes either in the planes or in the chains are associated with oxygen sites only. On the other hand, oxygen doping increases the number of holes both at the chain site 0 4 and the apical site O1 and introduces again superconductivity in oxygen-depleted Ca-123 samples. Further, Merz et al. (1998) find that superconductivity appears only if both the Cu02 planes and the Cu03 chains particularly the apex sites" - have an appropriate number of holes. As Ca doping increases only the number of holes of the planes, and Ca-123 samples with depleted oxygen content do not show any superconductivity at all -

164

E. KALDIS Table 12 Stoichiometryand superconductivityof tetragonal Y1 yCayBa2Cu30~,see also fig. 51" Ca contenty

O content x

0 0.09 0.10 0.14

6.18 6.29 6.29 6.30

Meissner (%) 0.02 0.03 0.5 16

Tc(90%) (K) 79 26 40 48

After data of B6ttger and Kaldis (1994). independently of the number of holes in the planes - they conclude that only a proper balance of Ca doping and O doping may lead to a Tc, m~x by establishing optimal hole numbers both on the Cu2 planes' and the apical sites. Further NEXAFS investigations on the unoccupied electronic structure of p-doped HTc superconductors with the apex site substituted by halogen atoms (F, C1) seem to support the above findings, provided that the homogeneity of the samples is proven (Nficker 1999). Also, in analogy to the above, in oriented single crystals of the ladder compounds (Sr, Ca,Y)14, Merz et al. (1998) could study spectral features corresponding to hole states on both the chain and ladder sites. Their results show that the holes reside mainly on the oxygens of the chains. The results of this work give a new charge-transfer picture and underline the importance o f the apex site for superconductivity. Merz et al. (1998) find that Ca-doped oxygen-depleted 123-O6 samples are not superconducting, in clear contrast to the findings of Bernhard and Tallon (1996) and Tallon et al. (1995). Whereas the earlier literature (e.g., Greaves and Slater 1989) reports superconductivity in such samples, more recent work (Hejtmanek et al. 1996) finds no superconductivity. The explanation for these discrepancies is probably the very small Meissner effect. B6ttger (1994) and B6ttger and Kaldis (1994) have found that whereas small steps appear in the magnetization curves, the Meissner fraction even for some Cay-123-O6.3 samples is very small, so that practically most such samples are not superconducting (fig. 51). Table 12 shows their findings. These strongly O-depleted samples are of course all tetragonal. Merz et al. (1998) can simply explain the unexpected decrease of To,ma× found earlier by Bernhard and Tallon (1996) and Tallon et al. (1995) with increasing Ca doping for samples with optimally doped planes. The increase of carriers with Ca doping must be counterbalanced with a decrease of the O doping. The latter reduces the holes on the apical site and decreases the To,max. On the contrary, increase of the O doping increases both the hole population on the apical site and on the planes. As the latter become overdoped the Tc is lowered. In conclusion, recent investigations of the 123 Ca overdoping reveal a much more complex picture consisting of alloying superconducting 123 with a non-superconducting substance (Ca-123 molecules with localized Ca-induced hole carriers) supporting once more physical phase separation.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30~.

165

Thus, physicalphase separation does not appear only at the onset o f superconductivity. In various forms it is also present in the oxygen-underdoped and -overdoped phases, as well as in the Ca-doped phase. It is probably an important characteristic o f HTc superconductivity.

9. Significant points of a doping-dependent phase diagram 123-0x (6.0 < x < 7.0) Starting from characteristic changes o f the X-ray lattice parameters cleared from possible sample irreproducibilities by the high-accuracy oxygen analysis, it is not difficult to find that also some structural, optical and magnetic properties o f 123-Ox show appreciable changes near these oxygen contents. Even more sensitive to some o f these oxygen contents appear to be the oxygen Raman phonons. The conclusion is that these compositions are significant points in the T-x phase diagram and may help to illuminate further the correlation between structural and physical properties. Figure 109 summarizes the appearance o f such changes vs. x in structural (average structure X R D and ND, and local structure EM, EXAFS), magnetic (under normal and high pressure), and optical (Raman) measurements for polycrystalline samples and single l"3J

' =,,; r !

Beye;rs89" I' .e_ l

I

Plakhty94 Zimmerm.99

'a° 2ao

0 e-

I_ ~ e"

Ill

T CAR+BAO

. T-O~I'

dTo/dP BAO 'a-, b-, V CAR I !c-, V, BAO

!~

I

Mllllllllll,,o _ I N +o II ao 8 1t.o I a°

E <

I I

1

l~j~l ~!

m

-I -m L

a- BAO b-, BAO

Nil [

a-,0,c,v oo

t~.l ~

I IUl

EXAFS li 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0 . . . .

i

. . . .

i

. . . .

i

. . . .

i

. . . .

I

. . . .

i

. . . .

i

. . . .

i

. . . .

Oxygen content x

Fig. 109. Schematic illustration of phase relationships, structural transitions and T~ anomalies appearing in the nonstoichiometric range of 123-O=. First row: EM investigationsof Beyers et al. (1989). Second row: XRD and NPD on poly- and single-crystalline samples, after Plakhty et al. (1994, 1995). Third row: Hard X-ray singlecrystal refinements of superstructures, after von Zimmermann et al. (1999). Fourth and fifth rows: T, under normal and higher pressures, measured in equilibrium samples. Sixth to tenth rows: phase relationships deduced from lattice parameters of slow|y cooled samples (preparation methods described in sect. 3.1.2.2). Eleventh row: EXAFS results after R6hler et al. (1997a,b, 1998).

166

E. KALDIS

crystals. The individual investigations have been discussed in detail throughout this review. Although the various phases and transitions depend on the method of preparation, five concentration regions include most observed anomalies. (a) The T - O transition appears in the range o f - 6 . 3 - 6 . 4 depending on the investigated samples and the method used. For CAR samples the onset of the transition appears at x = 6.3, supported by micro-Raman on individual crystallites. Deconvolution of the apex-phonon width indicates the coexistence of two phases. The I-M transition (onset x = 6.4) seems correlated to the structural transformation. (b) At the other end of the nonstoichiometric range (optimally doped and overdoped), changes of slope of the lattice parameters follow the onset of the optimally doped Tc plateau at x = 6.90. Also, changes of slope of the c-lattice parameter, ND and EXAFS manifest clearly the existence of a displacive transformation at x = 6.95. This transition separates the optimally doped phase of 123-Ox, which is its thermodynamically stablest form, from the phase-separated overdoped phase (as demonstrated from the splitting of the diamagnetic transition with onset x = 6.95). (c) In the intermediate range the chain superstructures appear. The 1D character of their chains may be important for the formation of stripes and the superconducting interactions in general. For this reason their investigation is of importance. Originally discovered by EM, the 2a0 and 3a0 have been studied now by XRD and ND and their structure is appreciably well known. Whereas the 2a0 is a 3D structure, the higher members are only 2D superstructures appearing as domains of the basal (a-b) plane. Higher superstructures (ortho-V and ortho-VIII) found with XRD are only combinations of the 2a0 and 3a0. Long-range interactions must exist, however, forming combinations up to eight chains. These superstructures are presently considered to be metastable. Therefore, as their appearance depends on a delicate balance of various interactions, it is expected that it will be strongly influenced by the history of the samples. Two models for the generation of superstructures have been used: The classical Gibbs thermodynamics predict singlephase regions separated by two-phase regions (miscibility gaps), and the cluster variation method (CVM) uses geometrical criteria to predict a combination of parent superstructures leading to sequences of empty-full chains, without separating twophase regions. The latter are of course difficult to prove experimentally with single crystals. Depending on the complexity of the phase diagram at high temperatures additional segregation phenomena are imposed during crystal growth. Up to now, the sequence of superstructures 2a0, ortho-V, ortho-VIII, and 3a0 have been found (single crystals near RT) without the separating two-phase regions. Lattice constants of samples resulting from one preparation method (BAO) do show horizontal parts, indicative of two phase regions, although no coexisting phases could be traced by XRD. Finally, the in-phase phonon shows steps near compositions where the nao superstructure series should appear. These steps seem to be separated by two-phase regions. The actual structure of these phases is not known. Also, it is not known if

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBa2Cu3O~

167

6,40 6,45 6,50 6,55 6,60 6,65 6,70 6,75 6,80 '''1

''1''''1''''1

....

I ' ' ' ' l ' ' ' ' l ' ' ' ' l

....

underdoped

A

E 452

•

o

2ao

450 C -~

448-

2ao+3~

@

O"

|

446-

w

u,

e- 4 4 4 O C

O 442ee~ ~-. 4 4 0 -

0

436-

O

"

~-~ 4 3 6 434-

°

~ o

';

~ O

3ao |

"~ e-

3a0+5a° 5ao •

o

i

•

;

--~ E

(5

72C

t-

II

D_ X

E

~ 432e-

.

~-

"O

e-

430 4286,40 6,45 6,50 6,55 6,60 6,65 6,70 6,75 6,80 oxygen content x Fig. 110. In-phase phonon frequency (Raman shifts) vs. oxygen content x for 6.40 <x < 6.80. The construction of the phase fields is discussed in sect. 5.5.2. After Kaldis et al. (2000b) with data of Poutakis et al. (1996) and Palles et al. (2000a).

these steps correspond to changes of structure or some other property, taking place at the nao oxygen contents. As shown in fig. 109, phase boundaries and two-phase regions consistently appear roughly in the ranges x ~ 6 . 6 , 6.7 and 6.8. As an example, we discuss here the tentative sequence of"dimpling-chain-superstructures" and construct a more detailed T - x structural phase diagram based on the room-temperature (RT) Raman measurements of the in-phase (02,03 vibrations tl c-axis) phonon frequency vs. oxygen stoichiometry, and complemented by the results of structural (RT and LT), and magnetic (LT) measurements. We note once more that most of the data leading directly.or indirectly to this phase diagram have been measured in large series of equilibrium samples (sect. 3.2.2) analyzed with high-accuracy oxygen volumetric analysis. The apex and in-phase phonon vs. x diagrams (~icro-Raman sect. 5.5) are shown in figs. 61, 64 and 65, and the in-phase phonon in expanded scale (x > 6.40) in figs. 110 and 111, respectively. The significant points of the phase diagram appear in the following oxygen contents referring to these figures:

168

E. K A L D I S 6,80 '

456

,

6,85

i

6,90

underdoped

7,00

optimally 3verdope cl doped ,o

454 E 0

6,95

i

- 456 - 454

452 450 e-

r0" 446-

5ao

2 U

5ao + 8ao

Q. .~

C 444O e"

O 442

..~

; 8

O. ,,... 440

I

|

8 - 438 2

8ao •

•

!

1

•

+ o co

|

•

•

•

r,, o. "D

- 442

~.

- 440

o

| •~'0 4 3 6 -

e

¢

434~ 432e" o

P-

~- A

428

- 430 - 428

6,80

6,85

6,90

6,95

7,00

oxygen content x Fig. 111. Same as fig. 110, but for 6 . 8 0 < x < 7 . 0 0 . After Kaldis et al. (2000b) with data o f Poulakis et al. (1996) and Palles et al. (2000a).

x = 6.20 Mesoscopic, RT

Micro-Raman: The Raman apex phonon frequency (figs. 61, 64), which - from the structure view point - is expected to be both sensitive to chain and plane changes, shows at this composition an abrupt softening with decreasing oxygen content indicating a transition inside the insulator regime. This is attributed to the AF transition at RT. The width changes also, indicating the coexistence in the whole 6.0 < x < 7.0 range of four different phases (fig. 63). Probably physical phase separation, as the X-ray investigations (cf. fig. 25) up to now do not show structural changes. It is possible that the appearance of the metallic clusters takes place at 6.20 and the percolation leading to the superconducting phase at 6.40. More compositions and synchrotron radiation investigations are needed to clarify the structural picture. x = 6.30 Macroscopic, RT Structure: In the range x = 6.30-6.40 the a- and b-lattice parameters show a widening (figs. 25a,b) due to the onset of the T --~ O transition at 6.30.

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTS IN YBa2Cu3Q

169

x = 6.40

Macroscopic, LT Superconductivity: The insulator-superconductor transition takes place near this composition probably as a result of the percolation of the hole doped clusters. At x--6.342 small amounts of the material become superconducting (fig. 27b). Superconductivity under pressure: A giant dTo/dP effect has been measured at RT (sect. 5.4) indicating an increased compressibility at the T - O transition due to the halffilled chains and strong oxygen-ordering effects. The effect disappears at LT where the mobility of oxygen freezes (fig. 58). Mesoscopic/microscopic, RT Raman: Both apex and in-phase phonons show smaller changes of slope, but much less abrupt and significant than at x = 6.20. Macroscopic, RT-LT Structure: The structure reacts at the onset of superconductivity with an abrupt decrease of the c-axis (sect. 3.2, figs. 12c, 25d, 25e) and the apical bond (fig. 14). x = 6.4-6.5

Mesoscopic/microscopic, RT Raman: The statistics of the microcrystallites indicate the existence of two steps of the in-phase phonon ( 0 2 - 0 3 ) frequency (one between 6.40-6.43 and another at 6.50) nearly overlapping with the range of the 2a0 chain superstructure (fig. 110 and sect. 5.5.2). Macroscopic, RT-LT Structure: Single crystal Rietveld refinement in bulk single crystals shows the onset of 2a0 at x ~ 6.35 (fig. 44, sect. 5.1.1). x = 6.5-6.6

Mesoscopic/microscopic, RT Raman: The statistics of the microcrystallites show two steps in the in-phase phonon ( 0 2 - 0 3 ) frequency (one at 6.50 and another between 6.57-6.60) where two in-phase "phases" coexist. In a nao series of superstructures this could be the range of the 2a0-3a0 miscibility gap (coexistence of the two superstructure phases) (fig. 110, sect. 5.5.2), the frequencies of each phase remaining constant with x as expected from the phase rule. x = 6.58-6.66

Mesoscopic/microscopic, RT Raman: in-phase phonon frequency indicates a single phase. The stoichiometry of the 3a0 superstructure is x = 6.66 (small asterisk in lower part of fig. 110). x = 6.66-6.75 Mesoscopic/microscopic, RT Raman: in-phase phonon frequency indicates a "two-phase" region, possibly (3a0 + 5a0), between the 3a0 and 5a0 compositions. x = 6.75-6.78

Macroscopic, LT, RT Superconductivity: The Tc-onset shows a step (fig. 2) and a miscibility gap.

170

E. KALDIS

Superconductivity under pressure: Another giant maximum of dTc/dP appears at this oxygen content (fig. 52) indicating an important transition or cross-over (sect. 5.4). Structure: The a- and b-parameters start to deviate from linearity (fig. 25a-c). Mesoscopic/microscopic LT, RT Raman: The apex phonon frequency becomes" independent of the oxygen content (fig. 61). The in-phase phonon indicates a phase boundary (transition). From stoichiometry it could be the boundary between the two-phase region (3a0 + 5a0), and the 5a0 superstructures (fig. 110), (sect. 5.4.2). x = 6.75-6.82

Mesoscopic/microscopic, RT Raman: In-phase phonon frequency shows a single phase behavior. The composition of the 5a0 superstructure would be x = 6.80 (fig. 111). x = 6.82

Macroscopic, RT Structure: Spikes in the unit cell volume, c-axis (and the other lattice parameters) of DO samples, figs. 30. Mesoscopic/microscopic, RT Raman: Onset of a "two-phase" region. From here on the stoichiometries of the nao superstructures do not follow exactly the sequence of the in-phase phases (fig. 111). x = 6.82-6.90

Mesoscopic/microscopic, RT, LT Raman: In-phase phonon frequency "two-phase" regions. The ideal stoichiometry of the 8a0 lies with 6.875 inside this range. In case of actual superstructures this would mean disorder: vacancies in the chains shifting the appearance of the superstructures at higher compositions (sect. 5.1.1). From the sequence, it could be a miscibility gap 5a0+8a0 (fig. 111). Local structure EXAFS (T =25 K): Y-O2,O3 spacing remains rather independent of oxygen doping in this range (fig. 77). x = 6.90

Macroscopic, LT Superconductivity: Onset optimally doped region, Tc,max plateau. Structure: Anomalies of unit cell volume and c-axis of DO samples (figs. 30). Mesoscopic/microscopic, RT, LT Raman: In-phase phonon frequency, transition from two-phase region to a single phase region. Analogy to a phase boundary between 5a0 + 8a0 and 8a0 superstructures (fig. 111). Local structure EXAFS (T = 25 K): Onset of stronger increase of dimpling (fig. 76). x = 6.92

Structure: Theoretical composition of a 13ao superstructure! x = 6.95

Macroscopic, RT, LT Superconductivity: Onset of overdoped region.

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN YBazCu30:,

171

Magnetism: Splitting diamagnetic transition (Tcl, To2), fig. 85. Structure RT and LT: c-axis and unit cell volume anomaly BAO, DO and CAR samples (XRD, ND; figs. 26c, 29t, 30). Dimpling (Cu2-O2,O3) discontinuity neutron diffraction (fig. 66a). Mesoscopic/microscopic, RT Raman RT: Abrupt softening in-phase phonon (fig. 79). Mesoscopic, RT Local structure EXAFS (T = 25 K): Dimpling abrupt increase (fig. 76) due to jump of the O2,O3 plane away from the Cu2 plane. Displaciue martensitic transformation (sect. 6.4). x = 6.95-6.97 Mesoscopic/microscopic, RT, LT Raman RT: In-phase phonon possibly indicates a "two-phase" region. Analogy with 8a0 +A, the overdoped phase (fig. 111). Local structure EXAFS (T=25 K): Dimpling remains constant in this doping range (fig. 76). x = 6.97 Macroscopic, RT Structure: Unit-cell volume and c-axis anomalies (XRD, fig. 30). Local structure EXAFS (T =25 K): Dimpling abruptly decreases due to shift of the O2,O3 plane towards the Cu2 plane (fig. 76). Spike in the DO lattice parameters. Mesoscopic, RT Raman: In-phase phonon indicates the stability region of the overdoped phase A (fig. ! 10). In the above, the compositions of superstructures have been calculated from the sequence nj_2+nj_l=nj for the superstructures 2a0, 3a0, 5a0, 8a0, . . . , nao with increasing O content, as has been proposed in the past (Kaldis et al. 1997b). In figs. 110 and 111 the asterisks in the lower part show oxygen contents for such superstructures. Above 5a0, shifts of the compositions start. The existence of higher chain superstructures has now been experimentally proven also for bulk single crystals (von Zimmermann et al. 1999) but with other sequences. As already mentioned, the overlap of the composition fields of (a) the experimentally found 2a0, 3ao, and the extrapolated 5a0, 8a0 (... nao) members of this sequence with (b) the compositions of in-phase phonon steps ("phases"), as well as the appearance between them of "two-phase steps", brought up the possible existence of "dimpling-chain" superstructures (sect. 5.1.1). If such superstructures do exist, it would be interesting to study their properties. Their characteristic lengths might be important for superconductivity. 9.1. A T - x phase diagram for 123-0x

Combining the above significant points of the phase diagrams of figs. 110, and 111 with elements of the structural phase diagram of yon Zimmermann et al. (1999) above RT, we can construct a schematic T-x phase diagram for 123-Ox, as shown in fig. 112.

172

E. KALDIS

50r"

i::iii!:i)i:i~i:ii!iil

451 401

351 301 25(

I--

201

_Z"

151 E

10( I

50 6,0 6,1 6,2 6,3 6,4 6,5 6,6 6,7 6,8 6,9 7,0

oxygen content x Fig. 112. Proposed T - x phase diagram for 123-O~. After Kaldis et al. (2000b).

We adopt the high-temperature phase boundary of T ~ O and the temperature of transformation of the superstructures to the orthorhombic basic ortho-I (O) structure from von Zimmermarm et al. (1999), but we use the sequence of superstructures indicated by the Raman in-phase data, shown in figs. 110 and 111. We obtain a sequence of single-phase superstructure regions with intermediate regions of phase separation (miscibility gaps) at mesoscopic/microscopic scale. Presently, due to lack of experimental investigations, a mechanism of the coupling of dimpling to the superstructure chains is not known. Theoretical investigations like those of Andersen et al. (1995) may shed more light on this problem (R6hler 2000b). Due to the small scale of fig. 112 the decrease of Tc in the overdoped range is not shown. Also the critical points of the phase-separated regions are kept rather flat due to lack of knowledge of their properties. It is interesting now to compare this diagram with that of La2CuO4+y (Wells et al. 1997), even if some details have changed in the meantime (fig. 113). Both phase diagrams show a normal state interrupted by a sequence of phase-separated regions. In between appear single-phase oxygen-intercalated structures with changing characteristic lengths. The superconducting states appear in these regions at lower temperatures. In both phase diagrams we have a sequence o f structural instabilities modulating the normal and superconducting states. It is possible that this is a generic element of the phase diagram of HTc cuprates.

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN Y B a 2 C u 3 0 ~

173

La2.:rSrxCuO4 Phase Diagram 500¸ 400 300 E

200

100 0.02

0.05

0.10 0. ! 5 Sr~'÷ Doping (x)

020

La2CuO4+yPhase Diagram 4® .... 3(K~ 20O 8. E

0 0 Stage 6

Stage 4 Stage 3 Stage 2

Oxygen Doping C~'and Staging #) Fig. 113. T-x phase diagram for La cuprates. After Wells et al. (1997). In the phase diagram of fig. l 13 several types of stripes have been found, following the ingenious discovery of stripes by Tranquada et al. (1995) and the concept of quantum wires (Bianconi and Missori 1994). Whereas for the La cuprates a folklore of stripes have been reported by many authors, up to very recently, this, type of phase separation has not been experimentally proven for 123. Very recently, however, two papers (Sharma et al. 2000, Mook et al. 2000) have presented evidence that stripes do exist also in 123-Ox (sect. 7.4).

174

E. KALDIS

Summary After five decades of semiconductor research in pursuit of the ideally homogeneous and perfect material, the picture of HTc materials emanates as intrinsically inhomogeneous. This review has tried to show some of these aspects: (A) There is a wealth of data showing the existence of important lattice effects and intrinsic inhomogeneities in 123-Ox.

(1) Five phase transitions exist: - The well-known T - O transition at x = 6.4, which according to Kartha et al. (1995) and Krumhansl (1992) is a first-order martensitic transformation. Also a neutron diffraction refinement (Radaelli et al. 1992) concludes that it is a first-order structural transition (sect. 3.2.1). The onset of superconductivity (I-M transition) takes place inside the composition range of the T - O transition (fig. 25d, e) and the latter triggers the former, as originally proposed by Cava et al. (1990). The ideas of a universal hole count for all cuprates to become superconducting (Tallon et al.) are doubtful for 123-Ox because their proof was based on questionable samples (sect. 5.3, 8.4). Recent structural data on very slowly cooled equilibrium samples (Conder et al. 1999; sect. 3.2.2) support clearly the abrupt change of the c-axis and apical bond found earlier by Cava et al. (1990). Charge transfer from the chains to the planes is coupled to this c-axis contraction. A giant dTJdP effect is measured at this composition which seems to be correlated with the oxygen ordering taking place at this transition (sect. 5.4). Raman investigations show another transition or cross-over at the N~el temperature (x=6.22). Deconvolution of the apex-phonon width shows the coexistence of four phases in the nonstoichiometric range of 123. It is possible that physical phase separation starts at x = 6.22 with the formation of hole-doped islands in an AF matrix (sect. 3.3.2), the percolation to the superconducting phase taking place at x ~ 6.40. - A fourth latent transition or cross-over takes place at x=6.75. At this composition the a- and b-lattice parameters deviate from linearity, and orthorhombicity, mimicking Tc, starts saturating (figs. 25a-c). It is possible, therefore, that this composition marks the start of the negative interactions leading to a decreasing T~ in the overdoped phase. Structurally, these changes are mainly in the planes, but have appreciable influence on the apex phonon frequency which for x > 6.75 becomes independent of doping (although its bond length is further decreasing). The most pronounced physical effect is the giant dTJdP increase (sect. 5.4), which is also connected with thermal expansion anomalies of the a-axis near T~ (sect. 5.4.2). - A fifth transition exists at the onset of the overdoped phase, x - 6.95. This is a displacive transformation (Kaldis et al. 1997b) followed by a decrease of orthorhombicity and a stepwise decrease of the b-axis (fig. 25b), but most importantly by an abrupt increase of dimpling (sect. 6). This transition is

OXYGENNONSTOICHIOMETRYAND LATTICEEFFECTSIN Y B a 2 C u 3 0 ~

175

the boundary of the phase-separated overdoped phase. This is supported by the splitting of the diamagnetic transition to Tc2,onset (almost independent of oxygen doping) and To1, and also by Raman measurements. The physical phase separation is probably due to overdoped islands in a matrix of the optimally doped phase. Thus, both ends of the superconducting field seem to bephase-separated (6.15 < x < 6.40, and 6.95 < x < 7.00). The Tc2,onsetappears with the same value in the Ca-doped 123 (see below). The transitions at x = 6.75 and 6.95 lead to changes parallel to the c-axis, but are induced by changes of orthorhombicity. The latter is triggered by displacement of 02,03 parallel to the c-axis and away from Cu2 (which moves out of the planes). (2) Several lattice distortions have been found in the underdoped and optimally

doped fields. - PDF (sect. 4.2) shows that in the local structure, shifts of 0.1 A of the oxygen sites appear along the c-axis in domains of~10 A. The question arises whether this is the result of the presence of polarons or some kind of stripes, as was shown to be the case for the La cuprates. - Electron diffraction shows correlated displacements of Cu2 and O1 in the a-b plane (sect. 4.3), leading to local perturbation of charge density roughly parallel to the c-axis, supporting the results of PDE These perturbations form a network of cells in the a-b plane with nanoscopic dimensions, comparable with the coherence length. The reason for these distortions is the strain field, which has been shown to exist also with the bond valence sum method in the 123-perovskite lattice (sect. 3.2.1.4). These results support the nanoscale phase separation model of Phillips and Jung (2001a,b) (sect. 4.3.1). - Y-EXAFS shows displacements of the 0 2 , 0 3 and Cu2 atoms parallel to the c-axis, inducing distortions of the dimpling. Lattice distortions appear also both at Tc and the spin-gap opening temperature. (B) A large number of experiments support the existence of several fields of phase separation. We have mentioned under (1) and (2) the physical phase separation at the two boundaries of the superconducting regime. In the underdoped range a kind of phase-rule obeying phase separation in mesoscopic scale can be considered, due to the existence of a series of in-phase phonon-frequency steps ("phases"), appearing at compositions where also the chain superstructures appear. This puts the question about the existence of dimpling-chain superstructures, where the chain-ordering phenomena would be correlated with the dimpling Raman shifts parallel to the c-axis (sect. 5.5.2). (figs. 110, 111). (C) A bag full of interesting and unexpected complexities is the case of Ca-doped 123-Ox. The structural data and the Tc change give the impression of classical overdoping: c-axis and apical bond increasing, To decreasing with Ca doping. A thorough inspection of the magnetization data of fully oxy.,genated samples shows that there also is a strong diamagnetic splitting and, therefore, phase separation following the pattern of the O-overdoped 123, and supported by the appearance of new Raman modes (sect. 8.2). Tcz,onset has the same value as for optimally O-doped 12306.92

176

E. KALDIS

and is almost independent of Ca doping, Td decreases strongly with Ca doping. The question about the local structure opens Pandora's box. Y-EXAFS sees, up to Reff ~ 5 A, almost no changes in the spectra and at higher Ca doping something like amorphisation. A model accommodating these difficulties (sect. 8.4) is the monodisperse dissolution of Ca-123 molecules in the lattice (R6hler et al. 1999b). This is expected to lead to localization of the holes introduced by Ca, the carriers screened by the locally distorted Y-123 cells surrounding the Ca-123. Then, the overdoping does not come from the holes doped by Ca but by the decreasing Y-123 contributions to the superconducting carriers with increasing Ca content. Not only may all of these be wrong in our picture of Ca-123. The assumption of linear dependence of O content on Ca% is also wrong (sect. 8.1). Up to 17% Ca very little oxygen is lost, not supporting the classical picture of substitution, except if the Ca holes are not available to the lattice. And last but not least, Ca-12306 which has been used to support the adoption of the parabolic dependence of Tc on the carrier concentration for 123 (Tallon et al. 1995) is not superconducting! 12306 doped with 5 different Ca compositions (0-22%) was found not to be superconducting (fig. 51) as discussed in sect. 8.4 (Merz et al. 1998). Summing up we can say that from the structural point of view the lattice strains are very important, leading to most of the local structure effects. The above is a list of experimental facts, which depending on the theoretical model can be interpreted differently. The ultimate goal of this review is to bring these facts to the attention of the reader (or remind some of them). Because whatever their interpretation may be, these experimental facts should not be overlooked.

Acknowledgments Many thanks are due to Professor J. R6hler for critically reading the manuscript, to Professors K.A. Milller, H. Keller, E. Liarokapis, and J. R6hler for many fruitful discussions, and to Professor E. Liarokapis and Drs G. B6ttger, K. Conder, P. Fischer, A.W. Hewat, J. Karpinski, I. Mangelshots-Loquet, D. Palles, Ch. Rossel, S. Rusiecki, H. Schwer, A. Shengelaya and D. Zech for longstanding collaborations and/or permission to use some yet unpublished data. For the discussion of the phase relationships of 123-Ox and the scouting for phase-separation phenomena the input of the Raman investigations as a function of nonstoichiometry was found to be an invaluable asset. My particular thanks go to Professor E. Liarokapis and his collaborators for these important contributions. Last but not least the technical support of E. Jilek, S. Mollet and A. Wisard over the years is gratefully acknowledged. Permissions for publication of figures and/or preprints and reprints have been received from Dr. N.H. Andersen, Prof. E Bridges, Dr. R Burlet, Prof. R. Cava, Prof. T. Egami, Dr. J. Etheridge, Dr. W.H. Fietz, Prof. L.H. Greene, Dr. J.D. Jorgensen, Dr. C. Meingast, Dr. S.B. Qadri, Prof. R.R Sharma, and Dr. J. Tallon. The author acknowledges this with pleasure.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30 x

177

The experimental work from the author's laboratory reported here has been supported in the past by the Swiss National Science Foundation (National Program of Superconductivity) and the ETH. The author wishes to aknowledge this crucial support.

References Ahn, B.T., T.M. Giir, R.A. Huggins, R. Beyers, E.M. Engler, EM. Grant, S.S.R Parkin, G. Lim, M.L. Ramirez, K.P. Roche, J.E. Vazquez, V.Y. Lee and R.D. Jacowitz, 1988, Physica C 153-155, 590-593. Aim, B.T., VY. Lee, R. Beyers, T.M. Giir and R.A. Huggins, 1990, Physica C 167, 529. Alario-Franco, M.A., C. Chaillout, J.J. Capponi, J. Chenavas and M. Marezio, 1988, Physica C 156, 455. Almasan, C.C., S.H. Hahn, B.W Lee, L.M. Paulius, M.B. Maple, B.W. Veal, J.W Downey, A.P Panlikas, Z. Fisk and J.E. Schirber, 1992, Phys. Rev. Lett. 69, 680. Andersen, N.H., B. Lebech and H.E Poulsen, 1990, in: High T0 Superconductor Materials, Proc. E-MRS Symposium, Strassbourg, France, eds H. Habermeier, E. Kaldis and J. Schoenes, J. LessCommon Met. 164-165, 124-131. Andersen, N.H., et al., 1996, Annual Progress Report Riso-R-933(EN) (Riso Nat. Laboratory, Denmark) p. 43. Andersen, O.K., A.I. Liechtenstein, O. Jepsen and E Paulsen, 1995, J. Phys. Chem. Solids 12, 1573. Appelman, E.H., L.R. Morss, A.M. Kini, U. Geiser, A. Umezawa, G.W Crabtree and K.D. Carlson, 1987, Inorg. Chem. 26, 3237. Aukrust, T., M.A. Novotny, EA. Rikvold and D.P Landau, 1990, Phys. Rev. B 41, 8772. Bartelt, N.C., T.L. Einstein and L.T. Wille, 1989, Phys. Rev. B 41, 10759. Batlogg, B., and V.J. Emery, 1996, Nature 382, 20. Bednorz, J.G., and K.A. Miiller, 1986, Z. Phys. B 64, 189. Beech, E, S. Miraglia, A. Santoro and R.S. Roth, 1987, Phys. Rev. B 35, 8778. Benischke, R., T. Weber, W.H. Fietz, J. Metzger, K. Grube, T. Wolf and H. Wtihl, 1992, Physica C 203, 293. Beno, M.A., L. Soderholm, C.W. Capone, D.G. Hinks, J.D. Jorgensen, J.D. Grace, I.K. Schuller, C.U. Segre and K. Zhang, 1987, Appl. Phys. Lett. 51, 57.

Bernhard, C., and J.L. Tallon, 1996, Phys. Rev. B 54, 10201 Beyers, R., and T.M. Shaw, 1989, The structure of YIBa2Cu307- 6 and its derivatives, in: Solid State Physics, Vol. 42, eds H. Ehrenreich and D. Turnbull (Academic Press, New York) pp. 135-312. Beyers, R., E.M. Engler, RM. Grant, S.S.R Parkin, G. Lim, M.L. Ramirez, K.E Roche, J.E. Vazquez, VY. Lee, R.D. Jacowitz, B.T. Ahn, T.M. Giir and R.A. Huggins, 1987, Mater. Res. Soc. Syrup. Proc. 99, 77. Beyers, R., B.T. Ahn, G. Gorman, VY. Lee, S.S.E Parkin, M.L. Ramirez, K.P. Roche, J.E. Vazquez, T.M. Giir and R.A. Hnggins, 1989, Nature 340, 619. Bianconi, A., and M. Missori, 1994, The coupling of a Wigner polaronic charge density wave with a Fermi liquid A possible pairing mechanism in HT c superconductors, in: Phase Separation in Cuprate Superconductors, eds E. Sigmund and K.A. Mfiller (Springer, Berlin). Bianconi, A., and N.L. Saini, eds, 1997, Proc. Int. Conf. on Stripes, Lattice Instabilities, and High-T~ Superconductivity, December 1996, Rome, Italy, J. Superconductivity 10, 267-466. Bianconi, A., and N.L. Saini, eds, 2000, Stripes and Related Phenomena, Proc. 2nd Int. Conf. on Stripes, Lattice Instabilities, and High-To Superconductivity, June 1998, Rome, Italy (Kluwer Academic/Plenum, Dordrecht/New York). Bianconi, A., A. Congiu Castellano, M. de Santis, R Rudolph, R Lagarde, A.M. Flank and A. Marcelli, 1987, Solid State Commun. 63, 1009. Bianconi, A., N.L. Saini, A. Lanzara, M. Missori, T. Rossetti, H. Oyanagi, H. Yamaguchi, K. Ota and T. Ito, 1996, Phys. Rev. Lett. 76, 3412. Bischof, R., E. Kaldis and I. Latsis, 1983, J. LessCommon Met. 94, 117. Bishop, A.R., 2000, lecture at Proc. 2nd Int. Conf. off Stripes, Lattice Instabilities, and High-T~ Superconductivity, June 1998, Rome, Italy. Bishop, A.R., R.L. Martin, K.A. Mtitler and Z. Tesanovic, 1989, Z. Phys. B 76, 17.

178

E. KALDIS

Booth, C.H., E Bridges, E.D. Bauer, G.G. Li, J.B. Boyce, T. Claeson, C.H. Chu and Q. Xiong, 1995, Phys. Rev. B 52, 15745. Booth, C.H., E Bridges, J.B. Boyce, T. Claeson, B.M. Lairson, R. Liang and D.A. Bonn, 1996, Phys. Rev. B 54, 9542. Bordet, E, C. Chaillout, T. Fournier, M. Marezio, E. Kaldis, J. Karpinski and E. Jilek, 1993, Phys. Rev. B 47, 3465 B6ttger, G., 1994, Diploma Thesis (University of Leipzig). B6ttger, G., and E. Kaldis, 1994, unpublished. B6ttger, G., 1. Mangelshots, Ch. Krfiger, E. Kaldis and E Fischer, 1996, J. Phys.: Condens. Matter 8, 8889. B6ttger, G., H. Schwer, E. Kaldis and K. Bente, 1997, Physica C 275, 198. Brown, I.D., 1978, Chem. Soc. Rev. 7, 359. Brown, I.D., 1989, J. Solid State Chem. 82, 122. Brown, I.D., 1991, J. Solid State Chem. 90, 155. Brown, I.D., and D. Andermatt, 1985, Acta Crystallogr. B 41,244. Brown, I.D., and R.D. Shannon, 1973, Acta Crystallogr. A 29, 266. Bucher, B., J. Karpinski, E. Kaldis and E Wachter, 1990, Pressure dependence of T~ and anisotropic features in the family Y2Ba4Cu6+nO14+,,, in: High Tc Superconductor Materials, Proc. E-MRS Symposium, Strassbourg, France, eds H. Habermeier, E. Kaldis and J. Schoenes, J. Less-Common Met. 164-165, 20-30. Bucher, B., E. Kaldis, S. Rusiecki and P. Wachter, 1991, in: Abstract Book, XIII AIRAPT Conference, Bangalore, India, October 1991. Bucher, B., E. Kaldis, S, Rusiecki and P Wachter, 1992, Pressure dependence of T~ in the family YBaCuO, in: Proc. XIII AIRAPT Conf., Bangalore, India, 1991 (Oxford University Press, New Delhi). Bucher, B., E. Kaldis, Ch. Krfiger and P. Wachter, 1996, Europhys. Lett. 34(5), 391. Bucldey, R.G., D.M. Pooke, J.L. Tallon, M.R. Presland, N.E. Flower, M.E Staines, H.L. Johnson, M. Meylan, G.V.M. Williams and M. Bowden, 1991, Physica C 174, 383. Burdett, J.K., 1992, Physica C 191, 282. Burdett, J.K., and G.V. Kulkarni, 1989, Phys. Rev. B 40, 8708. Burlet, E, J.Y. Henry and L.E Regnault, 2000, Physica C 296, 205. Butera, R.A., 1988, Phys. Rev. B 37, 5909. Capponi, J.J., C. Chaillout, A.W Hewat, E Lejay, M. Marezio, N. Nguyen, B. Raveau, J.L. Soubeyroux,

J.L. Tholence and R. Tournier, 1987, Europhys. Lett. 3, 1301. Cava, R.J., B. Batlogg, C.H. Chen, E.A. Rietman, S.M. Zahurak and D. Werder, 1987a, Phys. Rev. B 36, 5719. Cava, R.J., B. Batlogg, C.H. Chen, E.A. Rietman, S.M. Zahurak and D. Werder, 1987b, Nature 329, 423. Cava, R.J., B. Batlogg, S.A. Sunshine, T. Siegrist, et al., 1988a, Physica C 153-155, 560. Cava, R.J., B. Batlogg, K.M. Rabe, E.A. Rietman, RK. Gallagher and L.W. Rupp Jr, 1988b, Physica C 156, 523. Cava, R.J., A.W Hewat, E.A. Hewat, B. Batlogg, M. Marezio, K.M. Rabe, J.J. Krajewski, W.E Peck Jr and L.V~ Rupp Jr, 1990, Physica C 165, 419. Ceder, G., M. Asta, WC. Carter, M. Kraitchman, D. de Fontaine, M.E. Mann and M. Sluiter, 1990, Phys. Rev. B 41, 8698. Cedcr, G., R. McCol-mack and D. de Fontanie, 1991a, Phys. Rev. B 44, 2377. Ceder, G., M. Asta and D. de Fontaine, 1991b, Physica C 177, 106. Chaillout, C., and J.R Remeika, 1985, Solid State Commun. 56, 833. Chaillout, C., M.A. Alario-Franco, J.J. Capponi, J. Chenavas, E Strobe1 and M. Marezio, 1988, Solid State Commun. 65, 283. Chmaissem, 0., Y. Eckstein and C.G. Kuper, 2001, Phys. Rev. B 63, 174510. Claus, H., S. Yang, A.R Paulikas, J.W. Downey and B.W. Veal, 1990, Physica C 171, 205. Claus, H., M. Braun, A. Erb, K. R6hberg, B. Runtsch, H. Wtihl, G. Br/iuchle, E Schweiss, G. Mtiller-Vogt and H. v. L6hneysen, 1992a, Physica C 198, 42. Claus, H., U. Gebhardt, G. Linker, K. R6hberg, S. Riedling, J. Franz, T. Ishida, A. Erb, G. MtillerVogt and H. Wfihl, 1992b, Physica C 200, 271. Conder, K., 1999, private communication. Conder, K., 2000, Oxygen Exchange in the Superconductors of the YBaCuO Family, Habilitationsschrift (Department of Materials Sciences, ETH, Zurich). Conder, K., and E. Kaldis, 1989, J. Less-Common Met. 146, 205. Conder, K., and E. Kaldis, 1995, unpublished. Conder, K., and Ch. KNger, 1996, Physica C 269, 92. Conder, K., S. Rusiecki and E. Kaldis, 1989, Mater. Res. Bull. 24, 581. Conder, K., Wang Longmei, E. Boroch and E. Kaldis, 1991, Eur. J. Solid State Inorg. Chem. 28, 487. Conder, K., J. Karpinski, E. Kaldis, S. Rusiecki and E. Jilek, 1992, Physica C 196, 164.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30a Conder, K., E. Kaldis, M. Maciejewski, K.A. Mfiller and E.E Steigmeier, 1993, Physica C 213, 282. Conder, K., D. Zech, Ch. Krtiger, E. Kaldis, H. Keller, A.W. Hewat and E. Jilek, 1994a, Indications for a phase separation in YBa2CU3OT_x, in: Proc. Int. Workshop on Phase Separation, September 1993, eds E. Sigmund and K.A. Mtiller (Springer, Berlin) Part V, Phase Separation in Cuprate Superconductors, pp. 210-224. Conder, K., Ch. KNger, E. Kaldis, D. Zech and H. Keller, 1994b, Physica C 225, 13. Conder, K., H. Schwer, A. Shengelaya and E. Kaldis, 1999, to be published. Conradson, S.D., and I.D. Raistrick, 1989, Science 243, 1340. Dai, R, H.A. Monk and F. Dogan, 1998, Phys. Rev. Lett. 80, 1738. Darhmaoui, H., and J. Jtmg, 1996, Phys. Rev. B 53, 14621. Darhmaoui, H., and J. Jtmg, 1998, Phys. Rev. B 57, 8009. David, W.I.E, O. Moze, E Licci, E Bolzoni, R. Cywinski and S. Kilcoyne, 1989, Physica B 156-157, 884. de Fontaine, D., 1984, J. Phys. A 17, L713-L717. de Fontaine, D., L.T. Wille and S.C. Moss, 1987, Phys. Rev. B 36, 5709. de Fontaine, D., G. Ceder and M. Asta, 1990a, in: High T~ Superconductor Materials~ Proc. E-MRS Symposium, Strassbourg, France, eds H. Habermeier, E. Kaldis and J. Schoenes, J. LessCommon Met. 164-165, 108-123. de Fontaine, D., G. Ceder and M. Asta, 1990b, Nature 343, 544. de Fontaine, D., M. Asta, G. Ceder, R. McCormack and G. Van Tendeloo, 1992, Europhys. Lett. 19, 229-234. de Leeuw, D.M., W.A. Groen, L.E Feiner and E.E. Havinga, 1990, Physica C 166, 133. Deutscher, G., 1999, Nature 397, 410. Domanges, B., M. Hervieu and B. Raveau, 1993, Physica C 207, 65. D'yakonov, VP., I.M. Fita, G.G. Levchenko and VI. Markovich, 1996, Phys. Solid State 38, 458. Eckstein, Y., and C.G. Kuper, 1999, Physica B 284-288, 404. Egami, T., 1994, Local structm'al distortion: implication to the mechanism of HTc superconductivity, in: Materials and Crystallographic Aspects of HTo-Supereonductivity, NATO-ASI Series E Vol.

179

263, ed. E. Kaldis (Kluwer Academic Publishers, Dordrecht) pp. 45-64. Egami, T., and S.J.L. Billinge, 1996, Lattice effects in high-Tc superconductors, in: Physical Properties of High Temperature Superconductors, Vol. V, ed. D.M. Ginsberg (World Scientific, Singapore) pp. 1-109. Egami, T., Y. Petrov and D. Louca, 2000, J. Superconductivity 13, 709. Emerson, J.R, D.A. Wright, B.E Woodfield, J.E. Gordon, R.A. Fischer and N.E. Phillips, 1999, Phys. Rev. Lett. 82, 1546. Emery, VJ., S.A. Kivelson and H.Q. Lin, 1990, Phys. Rev. Lett. 64, 475. Emery, V.J., S.A. Kivelson and O. Zachar, 1997, Phys. Rev. B 56, 6120. Engler, E.M., et al., 1987, J. Am. Chem. Soc. 109, 2848. Etheridge, J., 1996, Philos. Mag. 73, 643. Eyring, L., 1979, The binary rare earth oxides, in: Handbook on the Physics and Chemistry of Rare Earths, Vol. 3, eds K.A. Gschneidner Jr and L. Eyring (North-Holland, Amsterdam) ch. 27. Farneth, V~E., R.K. Borida, E.M. McCarron III, M.K, Crawford and R.B. Flippen, 1988, Solid State Commun. 66, 953. Fietz, W.H., J. Metzger, T. Weber, K. Grube and H.A. Ludwig, 1994, AIP Conf. Proc. 309(Part I), 703. Fietz, W.H., R. Quenzel, H.A. Ludwig, K. Grube, S.I. Schlachter, EW. Hornumg, T. Wolf, A. Erb, M. Klfiser and G. Miiller-Vogt, 1996, Physica C 270, 258. Flemming, R.M., L.E Schneemeyer, P.K. Gallagher, B. Batlogg, L.W. Rupp and J.V. Waszczak, 1988, Phys. Rev, B 37, 7920. Francois, M., A. Junod, K. Yvon, A.W. Hewat, J.J. Capponi, P. Strobel, M. Marezio and P. Fischer, 1988, Solid State Commun. 66, 1117. Frank, EC., and J.H. van der Merwe, 1949, Proc. R. Soc. A 198, 205; 200, 125. Furrer, A., ed., 1998, Neutron Scattering in Layered Copper-Oxide Superconductors, Vol. 20 of Physics and Chemistry of Materials with Low-Dimensional Structures (Kluwer Academic Publishers, Dordrecht). Gallagher, P.K., H.M. O'Bryan, S.A. Sunshine and D.W. Murphy, 1987, Mater. Res. Bull. 22, 995. Gallaghel; P.K., G.S. Grader and H.M. O'Bryan, 1988, Mater. Res. Bull. 23, 1491.

180

E. KALDIS

Ginsberg, D.M., ed., 1990-95, Series: Physical Properties of High Temperature Superconductors (World Scientific, Singapore). Goodenough, J.B., and M. Lungo, 1970, in: Landol% Berstein Group II, Vol. 4a, eds K.-H. Hellwege and A.M. Hellwege (Springer, Berlin) ch. 3a. Gor'kov, L.R, and A.V. Sokol, 1987, JETP Lett. 46, 420. Graf, T., 1991, Ph.D. Thesis No. 2480 (University of Geneva, Imprimerie Nationale) p. 18. Graf, T., G. Triscone and J. M/filler, 1990, J. LessCommon Met. 159, 349. Greaves, C., and RR. Slater, 1989, Supercond. Sci. Technol. 2, 5. Greaves, C., and P.R. Slater, 199t, Physiea C 175, 172. Groen, W.A., D.M. de Leeuw and L.F. Feiner, 1990, Physica C 165, 55. Grybos, J., D. Hohlwein, Th. Zeiske, R. Solmtag, E Kubanek, K. Eichhom and Th. Wolf, 1994, Physica C 220, 138. Gutmann, M., S.J.L. Billinge, E.L. Brosha and G.H. Kwei, 2000, Phys. Rev. B 61, 11762. Haase, J., 2000, private commtmication (Physics Department, University of Stuttgart). Haase, J., N.J. Curro, R. Stern and C.R Slichter, 1998, Phys. Rev. Lett. 81, 1489. Harris, D.C., 1990, The Chemistry of Superconductor Materials (Noyes Publications, Park Ridge, NJ). Harris, D.C., M.E. Hills and T.A. Vanderah, 1987, J. Chem. Ed. 64,847. Hecker, N.E., H.A. Mook, J.A. Harvey, N.W. Hill, M.C. Moxon and J.A. Golovchenko, 1994, Phys. Rev. B 50, 16129. Hejtmanek, J., Z. Jirak, K. Knizek, M. Dlouha and S. Vratislav, 1996, Phys. Rev. B 54, 16226. Hewat, A.W., 1994, Neutron powder diffraction on the ILL high flux reactor and high T~ superconductors, in: Materials and Crystallographic Aspects of HTc-Superconductivity, ed. E. Kaldis, NATO ASI Series E, Vol. 263 (Kluwer Academic Publishers, Dordrecht) pp. 17 44. Hewat, A.W, E. Kaldis, S. Rusiecki, P. Fischer and E. Jilek, 1991, unpublished data. Hohlwein, D., 1994, Superstructures in 123 compounds, in: Materials and Crystallographic Aspects ofHTo-Superconductivity, ed. E. Kaldis, NATO ASI Series E, Vol. 263 (Kluwer Academic Publishers, Dordrecht) pp. 65 8i. Homes, C.C., T. Timusk, D.A. Bonn, R. Liang and W.N. Hardy, 1995, Can. J. Phys. 73, 663.

Hun, C.J., A. Manthiram, L. Rabenberg and J.B. Goodenough, 1990, J. Mater. Res. 5, 9. Howald, C., R Fournier and A. Kapitulnik, 2001, LANL cund-mat/0101251. Hudson, E.W., S.H. Pan, A.K. Gupta, K.W. Ng and J.C. Davis, 1999, Science 285, 88. Hume, D.N., and I.M. Kolthoff, 1944, Ind. Eng. Chem. Anal. Ed. 16, 103. Iliev, M., C. Thnmsen, V Hadjiev and M. Carduna, 1993, Plays. Rev. B 47, 12341. Inderhees, S.E., M.B. Salamon, N.D. Goldenfeld, J.R Rice, B.G. Pazol, D.M. Ginsberg, J.Z. Liu and G.W. Crabtree, 1988, Phys. Rev. Lett. 60, 1178. Iuoue, M., T. Takemori and T. Sakuda, 1987, Jpn. J. Appl. Phys. 26, L2015. Ishikawa, M., Y. Nakazawa, T. Takabatake, A. Kishi, R. Kato and A. Maesono, 1988a, Physica C 153 155, 1089. Ishikawa, M., Y. Nakazawa, T. Takabatake, A. Kishi, R. Kato and A. Maesono, 1988b, Solid State Commun. 66, 201. Jacobson, A.J., J.M. Newsam, D.C. Johnston, D.P. Goshorn, J.T. Lewandowski and M.S. Alvarez, 1989, Phys. Rev. B 39, 254. Janod, E., A. Janod, T. Graf, K.-Q. Wang, G. Triscone and J. Mtiller, 1993, Physica C 216, 129. Jensen, D.M., T. Fiig, N.H. Andersen and P.A. Lindghrd, 1997, Annual Progress Report, Solid State Physics, Riso-R-933(EN) (Riso Nat. Laboratory, Denmark) p. 44. Johnston, D.C., A.J. Jacobson, J.M. Newsam, J.T. Lewandowski, D.R Goshorn, D. Xie and W.B. Yelon, 1987, in: Chemistry of High-Temperature Superconductors, ACS Symposium, August 1987, eds D.L. Nelson, M.S. Whittingham and T.E Gemge, ACS Symposium Series, Vol. 351 (American Chemical Society, Washington, DC) p. 136. Jorgensen, J.D., B.W. Veal, V~K. Kwok, G.W. Crabtree, A. Umezawa, L.J. Nowicki and A.R Paulikas, 1987a, Phys. Rev. B 36, 5731. Jorgensen, J.D., M.A. Beno, D.G. Hinks, L. Soderholm, K.J. Volin, R.L. Hitterman, J.D. Grace, I.K. Schnller, C.U. Segre, K. Zhang and M.S. Kleefish, 1987b, Phys. Rev. B 36, 3608. Jorgensen, J.D., H. Shaked, D.G. Hinks, B.W. Veal, A.R Paulikas, L.J. Nowicki, G.W. Crabtree, H. Claus, W.K. Kwok, L.H. Nunez and H. Claus, 1988, Physica C 153 155, 578. Jorgensen, J.D., B.W. Veal, A.P. Paulikas, L.J. Nowicki, G.W. Crabtree, H. Claus and W.K. Kwok, 1990a, Phys. Rev. B 41, 1863.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu~O~ Jorgensen, J.D., S. Pei, R Lightfoot, H. Shi, A.R Paulikas and B.W. Veal, 1990b, Physica C 167, 571. Jorgensen, J.D., S. Pei, R Lightfoot, D.G. Hinks, B.W. Veal, B. Dabrowski, A.R Paulikas, R. Kleb and I.D. Brown, 1990c, Physica C 171, 64. Jtmg, J., B. Boyce, H. Yan, M. Abdelhadi, J. Skinta and T. Lemberger, 2000, J. Superconductivity 13, 753. Kaldis, E., ed., 1990, Solid State Chemistry of High-T~ Superconductors - Materials and Characterization, Eur. J. Solid State & Inorg. Chem. (Gauthier-Villars, Paris) 27. Kaldis, E., ed., 1994, "Materials and Crystallographic Aspects of HTc-Superconductivity" , NATO ASI Series E, Vol. 263 (Kluwer Academic Publishers, Dordrecht). Kaldis, E., 1997, Lattice instabilities, structural anomalies and phase transitions in Y1 yCayBa2Cu30, Phases, in: Ten Years after the Discovery, Workshop on High-Tc Superconductivity, August 1996, Delphi, Greece, eds E. Kaldis, E. Liarokapis and K.A. Mtiller, NATO ASI Series E, Vol. 343 (Kluwer Academic Publishers, Dordrecbt) pp. 411-446. Kaldis, E., and B. Fritzler, 1982, Valence and phase instabilites in TmSex crystals, in: Progress in Solid State Chemistry, Vol. 14, eds G.M. Rosenblatt and W.L. Worrell (Pergamon Press, Oxford) pp. 95-139. Kaldis, E., R Fischer, A.W. Hewat, E.A. Hewat, J. Karpinski and S. Rusiecki, 1989, Physica C 162, 524. Kaldis, E., J. Karpinski, S. Rusiecki, B. Bucher, K. Conder and E. Jilek, 1991, Physica C 185-189, 190. Kaldis, E., E. Liarokapis and K.A. Mtiller, eds, 1997a, Ten Years after the Discovery, Proc. Workshop on High-Tc Superconductivity 1996, NATO Advanced Study Institute, August 1996, Delphi, Greece, NATO ASI Series E, Vol. 343 (Kluwer Academic Publishers, Dordrecht). Kaldis, E., J. R6hler, E. Liarokapis, N. Poulakis, K. Conder and RW. Loeffen, 1997b, Phys. Rev. Lett. 79, 4894. Kaldis, E., E. Liarokapis, N. Poulakis, D. Palles and K. Conder, 2000a, in: Stripes and Related Phenomena, Proc. 2rid Int. Conf. on Stripes, Lattice Instabilities, and High-Tc Superconductivity, June 1998, Rome, Italy, eds A. Bianconi and N.L. Saini (Kluwer Academic/Plenum, Dordrecht/New York) p. 211. Kaldis, E., et al., 2000b, to be published.

i 81

K/ill, M., M. yon Zimmermann, N.H. Audersen, J. Madsen, T. Frello, O. Schmidt, H.E Poulsen, J.R. Sclmeider and Th. Wolf, 1999, unpublished (Riso Nat. Laboratory, Denmark). Karen, R, and A. Kjekshus, 1991, J. Solid State Chem. 94, 298. Karpinski, J., E. Kaldis, S. Rusiecki, B. Bucher and E. Jilek, 1988, Nature 336, 660. Karpinski, J., S. Rusiecki, E. Kaldis and E. Jilek, 1991, High oxygen pressure investigations in the Y2Ba4Cu6 +,,O~4+,, Family, in: Proc. HIP-3 Conference, Osaka, June 1991, ed. M. Koizttmi (Elsevier, Amsterdam). Karppinen, M., and H. Yamanchi, 1999, Philos. Mag. B 79, 343. Karppinen, M., A. Fukuoka, L. Niinist6 and H. Yamauchi, 1996, Supercond. Sci. Technol. 9, 121. Kartha, S., J.A. Krumhansl, J.P. Set'xmaand L.K. Wickham, 1995, Phys. Rev. B 52, 803. Keller, S.W., K.J. Leary, A.M. Stacy and J.N. Michaels, 1987, Mater. Lett. 5, 357. Khachaturyan, A.G., and J.W. Morris, 1988, Phys. Rev. Lett. 6l, 215. Khachaturyan, A.G., and J.W. Morris, 1990, Phys. Rev. Lett. 64, 1989. Khachaturyan, A.G., S.V. Semenovskaya and J.W. Morris, 1988, Phys. Rev. B 37, 2243. Khomskii, D., and EV Kusmartsev, 1992, Phys. Rev. B 52, 803. Kimura, H., H. Oyanagi, T. Terashima, H. Yamaguchi, Y. Bando and J. Mizuld, 1993, Jpn. J. Appl. Phys. 32-2, 584. Kohn, W., T.K. Lee and Y.R. Lin-Liu, 1982, Phys. Rev. B 25, 3557. Kraut, O., C. Meingast, G. Br/iuchle, H. Claus, A. Erb, G. Mtiller-Vogt and H. Wtitfl, 1993, Physica C 205, 139. Kr/iger, Ch., K. Conder and E. Kaldis, 1993, Physica C 213, 219. Krfiger, Ch., H. Schwer, J. Karpinski, K. Conder, E. Kaldis, C. Rossel and M. Maciejewski, 1994, Z. Anorg. Allg. Chemie 620, 1932. Krfiger, Ch., K. Conder, H. Schwer and E. Kaldis, 1997, J. Solid State Chem. 134, 356. Krumhansl, J.A., 1992, Fine scale mesostructures in superconducting and other materials, in: Lattice Effects in High-To Superconductors, eds Y. Bar-Yam, T. Egami, J. Mustre-de Leon and A.R. Bishop (World Scientific, Singapore) pp. 503-516.

182

E. KALDIS

Kwei, G.H., A.C. Larson, WL. Huts and J.L. Smith, 1990, Physica C 169, 217. Lavrov, A.N., 1992, Phys. Lett. A 168, 71. Liarokapis, E., 1997, Raman scattering in YBCO compounds, in: Ten Years after the Discovery, Workshop on High-Tc Superconductivity, August 1996, Delphi, Greece, eds E. Kaldis, E. Liarokapis and K.A. Mfiller, NATO ASI Series E, Vol. 343 (Kluwer Academic Publishers, Dordrecht) pp. 447-468. Liarokapis, E., 1999, Physica C 317-318, 270. Liarokapis, E., 2000, Physiea C 341 348, 2185. Liarokapis, E., et al., 2000, to be published. Lindemer, T.B., C.R. Hubbard and J. Brynestad, 1990, Physica C 167, 312. Loram, J.W, J.R. Cooper and K.A. Mirza, 1991, Supercond. Sci. Technol. 4, $391. Louea, D., G.H. Kwei, B. Dabrowski and Z. Bukowski, 1999, Phys. Rev. B 60, 7558. Macejewski, M., A. Baiker, K. Conder, Ch. Kffdger, J. Karpinski and E. Kaldis, 1994, Physica C 227, 343. MacFarlane, R.M., H.J. Rosen, E.M. Engler, R.D. Jacowitz and V.Y. Lee, 1988, Phys. Rev. B 38, 284. Maletta, H., E. P6rscbke, B. Rupp and E Meuffels, 1989, Z. Phys. B 77, 181. Marezio, M., and C. Chaillout, 1994, A classification of the copper oxide superconductors and the relationship between the Cu valence and the superconducting properties, in: Materials and Crystallographic Aspects of riTe-Superconductivity, ed. E. Kaldis, NATO ASI Series E, Vol. 263 (Kluwer Academic Publishers, Dordrecht) pp. 3-16. McQueeney, R.J., Y. Petrov, T. Egami, M. Yethiraj, G. Shirane and Y. Endoh, 1999, Phys. Rev. Lett. 82, 628. Medvedeva, I.V., Y.S. Bersenev, B.A. Gizhevsky, N.M. Chebotaev, S.V. Naumov and G.B. Demishev, 1990, Z. Phys. B 81, 311. Meingast, C., O. Kraut, T. Wolf, H. WOhl, A. Erb and G. Mtiller-Vogt, 1991, Phys. Rev. Lett. 67, 1634. Merz, M., N. Nticker, P. Schweiss, S. Schuppler, C.T. Chen, V. Chakarian, J. Freeland, Y.U. Idzverda, M. Klhser, G. Mfiller-Vogt and Th. Wolf, 1998, Phys. Rev. Lett. 80, 5192. Metzger, J., T. Weber, W.H. Fietz, K. Grube, H.A. Ludwig, T. Wolf and H. W/lhl, 1993, Physica C 214, 371. Miceli, EE, J.M. Tarascon, L.H. Greene, P Barboux, EJ. Rotella and J.D. Jorgensen, 1988, Phys. Rev. B 37, 5932.

Mihailovic, D., and K.A. Mfiller, 1997, in: Ten Years after the Discovery, Workshop on High-Tc Superconductivity, August 1996, Delphi, Greece, eds E. Kaldis, E. Liarokapis and K.A. MiJller, NATO ASI Series E, Vol. 343 (Kluwer Academic Publishers, Dordreeht). Mihailovic, D., G. Ruani, E. Kaldis and K.A. Mtiller, eds, 1995, Anharmonic Properties of High-T c Cuprates (World Scientific, Singapore). Moler, K.A., et al., 1997, Phys. Rev. B 55, 3954. Monod, E, M. Ribault, E D'Yvoire, J. Jegoudez, G. Collin and A. Revcolevschi, 1987, J. Phys. (Paris) 48, 1369. Monster, D., P.-A. Lindghrd and N.H. Andersen, 1999, unpublished. Mook, H.A., and E Dogan, 1999, Nature 401, 145. Mook, H.A., Y. Yethiraj, G. Aeppli, T.E. Mason and T. Armstrong, 1993, Phys. Rev. Lett. 70, 3490. Mook, H.A., et al., 1998, Nature 395, 580. Mook, H.A., P. Dai, E Dogan and R.D. Hunt, 2000, Nature 404, 729. Mfiller, K.A., 1994, Recipe for high-T~ transitionmetal chalcogenides, in: Phase Separation in Cuprate Superconductors, eds E. Sigmund and K.A. Mtiller (Springer, Berlin) pp. 1-11. Mfiller, K.A., and G. Benedek, 1992, Phase Separation in Cuprate Superconductors (World Scientific, Singapore). Mfiller, K.A., K.W. Blazey, J.G. Bednorz and M. Takashige, 1987, Physica B 148, 149. Murata, K., Y. Honda, H. Oyanagi, Y. Nishihara, H. Ihara, N. Terada, R. Sugise, M. Hirabayashi, M. Tokumoto and 3(. Kimura, 1989, MRS Int. Meeting on Advanced Materials, 6, 681. Murch, G.E., 1980, Atomic Diffusion Theory in Highly Defective Solids, Vol. 6 of Diffusion and Defect Monograph Series (Trans Teeh, Aedersmannsdorf). Mustre-de Leon, J., S.D. Conradson, J. Batisfic and A.R. Bishop, 1990, Phys. Rev. Lett. 65, 1678. Mustre-de Leon, J., J. Batistic, A.R. Bishop, S.D. Conradson and S.A. Trugman, 1992, Phys. Rev. Lett. 68, 3236. Mustre-de Leon, J., J. Batistic, A.R. Bishop, S.D. Conradson and I.D. Raistrick, 1993, Phys. Rev. B 47, 12322. Nakazawa, Y., 1991, Ph.D. Thesis (Prof. M. Ishikawa; Institute of Solid State Physics, University of Tokyo). Nakazawa, Y., and M. Ishikawa, 1989, Physica C 162-164, 83.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS IN YBa2Cu30, Nakazawa, Y., and M. Ishikawa, 1991, private communication. Nakazawa, Y., J. Takeya and M. Ishikawa, 1994, Physica C 225, 71. Narlikar, A., ed., since 1990, Series: Studies of High Temperature Superconductors (Nova Science Publisher, New York). Nelmes, R.J., J.S. Loveday, E. Kaldis and J. Karpinski, 1990, Physica C 172, 311. Nticker, N., 1999, Seminar lecture, Seminar for Superconductivity, 2 February 1999, Laboratorium Ftir Festk6rperphysik, ETH, Zurich. Nticker, N., J. Fink, B. Renker, D. Ewert, C. Politis, P.J.W. Weijs and J.C. Fuggle, 1987, Z. Plays. B 67, 9. Nficker, N., J. Fink, J.C. Fuggle, EJ. Durham and W.M. Temmerman, 1988, Phys. Rev B 37, 5158. Palles, D., 2000, Ph.D. Thesis (Department of Physics, National Technical University, Athens). Palles, D., N. Poulakis, E. Liarokapis, K. Conder, E. Kaldis and K.A. Mfiller, 1996, Phys. Rev. B 54, 6721. Palles, D., N. Poulakis, E. Liarokapis, K. Conder and E. Kaldis, 2000a, Physica C 341-348, 2163. Palles, D., D. Lampakis, G. B6ttger, E. Liarokapis and E. Kaldis, 2000b, Physica C 341-348, 2149. Pan, S.H., E.W. Hudson, K.M. Lang, H. Elsaki, S. Uchida and J.C. Davis, 2000, Nature 403, 746. Panling, L., 1929, J. Am. Chem. Soc. 51, 1010. Phillips, J.C., 1987, Phys. Rev. Lett. 59, 1856. Phillips, J.C., 1990, Phys. Rev. B 41, 8968. Phillips, J.C., 1991, Phys. Rev. B 43, 11415. Phillips, J.C., 1999, Philos. Mag. B 79, 1477. Phillips, J.C., 2000, Philos. Mag. B 80, 1773. Phillips, J.C., 2001, Philos. Mag. B 81, 35. Phillips, J.C., and J. Jtmg, 2001a, Phys. Rev. B, submitted. LANL cond-mat/0103167. Phillips, J.C., and J. Jung, 2001b, Philos. Mag. B, in press. Pickett, W.E., 1995, Aalharmonic oxygen displacements in La 2 x(Sr,Ba).~CuO4 planes and in YBaaCu307 chains, in: Proc. Int. Workshop on Anharmonic Properties of High-To Cuprates, Bled, Slovenia, September 1994, eds D. Mihailovic, G. Ruani, E. Kaldis and K.A. Mfiller (World Scientific, Singapore) pp. 159-170. Pickett, W.E., 1997, Physica C 289, 51-62. Pintschovius, L., and W. Reichardt, 1998, Phonon dispersion and phonon density states in copperoxide superconductors, in: Neutron Scattering in Layered Copper-Oxide Superconductors, ed.

183

A. Furrer, Vol. 20 of Physics and Chemistry of Materials with Low-Dimensional Structures (Khiwer Academic Publishers, Dordrecht). Plakhty, V., A. Stratilatov, Yu. Chernenkov, V. Fedorov, S.K. Sinha, C.L. Loong, G. Gaulin, M. Vlasov and S. Moshkin, 1992, Solid State Commun. 84, 639. Plakhty, V, B. Kviatkovsky, A. Stratilatov, Yu. Chernenkov, R Burlet, J.Y. Henry, C. Martin, E. Ressouche, J. Schweizer, F. Yakhou, E. Elkaim and J.R Lamiat, 1994, Physica C 235-240, 867-868. Plakhty, V., P. Burlet and J.Y. Henry, 1995, Phys. Lett. A 198, 256. Poulakis, N., D. Palles, E. Liarokapis, K. Conder, E. Kaldis and K.A. Mfiller, 1996, Phys. Rev. B 53, R534. Poulsen, H.F., M. yon Zimmermann, J.R. Schneider, N.H. Andersen, E Schleger, J. Madsen, R. Hadfield, H. Casalta, R. Liang, R Dosanjh and W.N. Hardy, 1996, Phys. Rev. B 53, 15335. Presland, M.R., J.L. Tallon, R.G. Buckley, R.S. Liu and N.E. Flower, 1991, Physica C 176, 95. Pyka, N., W. Reichardt, L. Pintschovius, S.L. Chaplot, P. Schweiss, A. Erb and G. Miiller-Vogt, 1993, Phys. Rev. B 48, 7746. Qadri, S.B., E.F. Skelton, M.S. Osofsky, VM. Browning, J.Z. Hu, ER. Bronssard, M.E. Reeves and W. Prusseit, 1997, Physica C 282-287, 93. Radaelli, RG., 1998, Structural anomalies, oxygen ordering and superconductivity in YBa2Cu306 + x, in: Neutron Scattering in Layered Coppm:Oxide Superconductors, ed. A. Fnrrer, Vol. 20 of Physics and Chemistry of Materials with Low-Dimensional Structures (Kluwer Academic Publishers, Dordrecht). Radaelli, PG., C.U. Segre, D.G. Hinks and J.D. Jorgensen, 1992, Phys. Rev. B 45, 4923. Ranninger, J., and U. Thibblin, 1992, Phys. Rev. B 45, 7730. Raveau, B., M. Huve, A. Maignan, M. Hervieu, C. Michel, B. Domanges and C. Martin, 1993, Physica C 299, 163. Reichardt, W., N. Pyka, L. Pintschovius, B. Hennion and G. Collin, 1989, Physica C 162-164, 464. Reyes-Gasga, J., T. Krekels, G. Van Tendeloo, J. Van Landuyt, S. Amelinckx, W.H.M. Bruggink and H. Verweij,,1989, Physica C 159, 831. R6hler, J., 1994, Evidence from EXAFS for an axial oxygen centered lattice instability in YBa~Cu307 _ ~, in: Materials and Crystallographic Aspects of HTc-Superconductivity, ed. E. Kaldis,

184

E. KALDIS

NATO-ASI Series E, Vol. 263 (Kluwer Academic Publishers, Dordrecht) pp. 353-372. R6hler, J., 2000a, Physica C 341 348, 2151. LANL cond-mat0003116. R6hler, J., 2000b, Physica B 284-288, 1041-1042. LANL cond-mat/9906432. R6hler, J., and R. Crfisemann, 1995, Multiple scattering analysis of Cu-K EXAFS in Bi2Srl.5CaLsCu208, in: Proc. Int. Workshop on Anharmonic Properties of High-T c Cuprates, Bled, Slovenia, September 1994, eds D. Mihailovic, G. Ruani, E. Kaldis and K.A. Miiller (World Scientific, Singapore) pp. 86-94. R6hler, J., A. Larish and R. Schiifer, 1992, Physica C 191,57. R6hler, J., RV~ Loeffen, S. Mtillender, K. Conder and E. Kaldis, 1997a, Local structure studies of the underdope~overdoped transition in YBa2Cu3Ox, in: Ten Years after the Discovery, Proc. Workshop on High-Tc Superconductivity 1996, eds E. Kaldis, E. Liarokapis and K.A. Miiller, NATO ASI Series E, Vol. 343 (Kluwer Academic Publishers, Dordrecht) pp. 469-502. R6hler, J., P.W. Loeffen, K. Conder and E. Kaldis, 1997b, Physica C 282 287, 182-185. R6hler, J., et al., 1998, unpublished. R6hler, J., E. Rose, J.B. Philip, T. Lipinski, G. B6ttger and E. Kaldis, t999a, Annual Report (HASYLAB). R6hler, J., C. Friedrich, T. Granzow, E. Kaldis and G. B6ttger, 1999b, Phase separation in overdoped YI 08Ca0 0.2Cu306.96-6.98,in: HTS99 Miami, January 1999, eds S.E. Barnes, J. Ashkenasi, J.L. Cohn and E Zuo, AIP Conf. Proc. 483, 320-323. LANL condmat/9902030. Rossel, C., et al., 1999, unpublished results (IBM Zurich Research Laboratory). Rothman, S.J., J.L. Routbort and J.E. Baker, 1989, Phys. Rev. B 40, 8852. Rothman, S.J., J.L. Routbort, U. Welp and J.E. Baker, 1991, Phys. Rev. B 44, 2326. Rusiecki, S., and E. Kaldis, 1991, unpublished work (ETH, Zurich). Rusiecki, S., B. Bucher, E. Kaldis, E. Jilek, J. Karpinski, C. Rossel, B. Piimpin, H. Keller, W. Ktindig, T. Krekels and G. Van Tendeloo, 1990, in: High Tc Superconductor Materials, Proc. E-MRS Symposium, Strassbourg, France, eds H. Habermeier, E. Kaldis and J. Schoenes, J. LessCommon Met. 164-165, 31 38.

Salomons, E., and D. de Fontaine, 1990, Phys. Rev. B 41, 11159. Sanchez, J.M., E Ducastelle and D. Gratias, 1984, Physica A 128, 334. Schilling, J.S., and S. Klotz, 1992, in: Physical Properties of High Temperature Superconductors, Vol. Ill, ed. D.M. Ginsberg (World Scientific, Singapore) p. 59. Schlachter, S.I., W.H. Fietz, K. Grube, Th. Wolf, B. Obst, R Schweiss and M. K1/iser, 1999, Physica C 328, 1-13. Schleger, E, W.N. Hardy and H. Casalta, 1994, Phys. Rev. B 49, 514. Schleger, E, R. Hadfield, H. Casalta, N.H. Andersen, H.E Poulsen, M. von Zimmermann, J.R. Schneider, R. Liang, P. Dosanjh and W.N. Hardy, 1995a, Phys. Rev. Lett. 74, 1446. Schleger, P., H. Casalta, R. Hadfield, H.E Poulsen, M. von Zimmermann, N.H. Andersen, J.R. Schneider, R. Liang, P. Dosanjh and W.N. Hardy, 1995b, Physica C 241, 103. Schmalzried, H., 1981, Solid State Reactions (Verlag Chemie, Weinheim). Schweiss, R, W. Reichardt, M. Braden, G. Collin, G. Heger, H. Claus and A. Erb, 1994, Phys. Rev. B 49, 1387. Schwer, H., J. Karpinski, K. Conder, Ch. Krfiger, E. Kaldis and C. Rossel, 1993a, 249. Schwer, H., E. Kaldis, J. Karpinski and C. Rossel, 1993b, Physica C 211, 165. Semenovskaya, S.V, and A.G. Khachaturyan, e.g.: 1993, Physica D 66, 205. Shafer, M.W., and T. Penney, 1990, Hole, To relationships in superconducting oxides, in: Solid State Chemistry of High-To Superconductors Materials and Characterization, ed. E. Kaldis, Eur. J. Solid State & Inorg. Chem. (Ganthier-Villars, Paris) 27, 191-220. Sharer, M.W., R.A. De Groot, M.M. Plechaty and G.J. Scilla, I988, Physica C 153 155, 836. Shaked, H., J.D. Jorgensen, J. Faber, D.G. Hinks and B. Dabrowski, 1989, Phys. Rev. B 39, 7363. Sharma, R.E, EJ. Rorella, J.D. Jorgensen and L.E. Rehn, 1991, Physica C 174, 409. Sharma, R.R, S.B. Ogale, Z.H. Zhang, J.R. Liu, W.K. Chu, B. Veal, A. Paulikas, H. Zheng and T. Venkatesan, 2000, Nature 404, 736. Shoenberg, D., 1940, Proc. R. Soc. London Set. A 175, 49.

OXYGEN NONSTOICHIOMETRY AND LATTICE EFFECTS 1N YBaaCu30~ Sigmund, E., and K.A. Miiller, eds, 1994, Phase Separation in Cuprate Superconductors (Springer, Berlin). Stauffer, D., and A. Aharony, 1995, Perkotationstheorie: eine Einfilhrnng (VCH, Weinheim, Germany) p. 18. Stern, E.A., M. Qian, Y. Yacoby, S.M. Heald and H. Maeda, 1992, Unusual copper environment in YBa2Cu3OT_ ~ superconductors as found by XAFS, in: Proc. Conf. Lattice Effects in High To Superconductors, eds Y. Bar-Yam, T. Egami, J. Mustre-de Leon and A.R. Bishop (World Scientific, Singapore) p. 51. Stern, E.A., M. Qian, Y. Yacoby, S.M. Heald and H. Maeda, 1993, Physica C 209,331. Sterne, R, and L.T. Wille, 1989, Physica C 162 164, 223. Straube, E., D. Hohlwein and E Kubanek, 1998, Physica C 295, 1. Sullivan, J.D., R Bordet, M. Marezio, K. Takenaka and S. Uchida, 1993, Phys. Rev. B 48, 10638. Swaminathan, K., J. Janaki, G.V.N. Rao, O.M. Sreedharan and T.S. Radhakrishnan, 1988, Mater. Lett. 6, 265. Takayama-Muromachi, E., Y. Uchida, M. Ishi, T. Tanaka and K. Kato, 1987, Jpn. J. Appl. Phys. 26, Ll156. Tallon, J.L., 1990, Physica C 168, 85. Tallon, J.L., and J.R. Cooper, 1993, in: Advances in Superconductivity, Vol. V, eds Y Bando and H. Yamaguchi (Springer, Tokyo) p. 339. Tallon, J.L., C. Bernhard, H. Shaked, R.L. Hitterman and J.D. Jorgensen, 1995, Phys. Rev. B 51, 12911. Tarascon, J.M., WR. McKimlon, L.H. Greene, G.W. Hull, B.G. Bagley, E.M. Vogel and Y. La Page, 1987a, Proc. 1987 MRS Spring Meeting, April 1987, eds D.V Gupser and M. Schluter (Materials Research Society, Pittsburg, PA) Vol. EA-11, p. 65. Tarascon, J.M., R Barboux, B.G. Bagley, L.H. Greene, W.R. McKinnon and G.W. Hull, 1987b, in: Chemistry of High-Temperature Superconductors, ACS Symposium, August 1987, eds D.L. Nelson, M.S. Whittingham and T.E George, ACS Symposium Series, Vol. 351 (American Chemical Society, Washington, DC) p. 198. Tetenbaum, M., B. Tani, B. Czech and M. Blander, 1989, Physica C 158, 377. Thomsen, C., 1991, in: Light Scattering in Solids, Vol. VI, eds M. Cardona and G. Gfintherodt (Springer, Berlin).

185

Thomsen, C., and M. Cardona, 1993, Phys. Rev. B 47, 12320. Thomsen, C., and M. Cardona, 1994, in: Physical Properties of High Temperature Superconductors, Vol. 5, ed. D.M. Ginsberg (World Scientific, Singapore). Tokura, Y., J.B. Torrance, T.C. Huang and A.I. Nazzal, 1988, Phys. Rev. B 38, 7156. Torrance, J.B., and R.M. Metzger, 1989, Phys. Rev. Lett. 63, 1515. Torrance, J.B., Y. Tokura, A.I. Nazzal, A. Benzike, T.C. Huang and S.S.R Parkin, 1988, Phys. Rev. Lett. 61, 1127. Tranquada, J.M., 1998, Physica B 241-243. Tranquada, J.M., B.J. Sternlieb, J.D. Axe, Y. Nakamm'a and S. Uchida, 1995, Nature 375, 561. Uchida, S., 1997, Physica C 282 287, 12. van der Merwe, J.H., 1963, J. Appl. Phys. 34, 117, 123. Van Tendeloo, G., and S. Amelinckx, 1990, in: High To Superconductor Materials, Proc. E-MRS Symposium, Strassbourg, France, eds H. Habermeier, E. Kaldis and J. Schoenes, J. LessCommon Met. 164 165, 92. Van Tendeloo, G., H.W. Zandbergen and S. Amelinckx, 1987, Solid State Commun. 63, 603. Veal, B.W., H. You, A.R Panlikas, H. Shi, Y. Fang and J.W. Downey, 1990a, Phys. Rev. 42, 4770. Veal, B.W., A.R Paulikas, Hoydoo You, H. Shi, Y. Fang and J.W. Downey, 1990b, Phys. Rev. B 42, 6305. von Zimmermann, M., T. Frello, N.H. Andersen, J. Madsen, M. K/ill, O. Schmidt, T. Niem611er, J.R. Schneider, H.E Poulsen, Th. Wolf, R. Liang, R Dosanjh and W.N. Hardy, 1999, preprint LANL cond-matter/9906251, submitted to Phys. Rev. B, April 7th, 1999 (to be published in revised form). Wada, T., H. Yamanchi and S. Tanaka, 1992, J. Am. Chem. Soc. 75, 1705. Wang, J., I. Monot, M. Hervieu, J. Provost and G. Descardin, 1996, Supercond. Sci. Tectmol. 9, 69. Weber, T., R. Benischke, W.H. Fietz, J. Metzger and T. Wolf, 1992, in: Proceedings DPG TTll.19, Spring Meeting Deutsche Physikalische Gesellschaft, Regensburg, Germany. Wells, B.O., Y.S. Lee, M.A. Kastner, R.J. Christianson, EC. Chou and R.J. Birgeneau, 1997, in: Ten Years after the Discovery, Workshop on High-Tc Supercoilductivity, August 1996, Delphi, Greece, eds E. Kaldis, E. Liarokapis and K.A. Miiller, NATO ASl Series E, Vol. 343 (Kluwer Academic Publishers, Dordrecht) pp. 349-370.

186

E. KALDIS

Welp, U., M. Grimsditch, S. Fleshier, W. Nessler, J. Downey, G.W. Crabtree and J. Guimpel, 1992, Phys. Rev. Lett. 69, 2130. Werder, D.J., C.H. Chen, R.J. Cava and B. Batlogg, 1988, Phys. Rev. B 38, 5130. Whangbo, M.-H., mad C.C. Torrardi, 1990, Science 249, 1143. Widder, K., et al., 1996, Physica C 267, 254. Wille, L.T., A. Berera and D. de Fontaine, 1988, Phys. Rev. Lett. 60, 1065. Woug-Ng, W., EW. Gayle, D.L. Kaiser, S.E Watkins and ER. Fronczek, 1990, Phys. Rev. B 41, 4220. Wu, W.K., J.R. Ashburn, C.J. Torng, P.H. Hot, R.L. Meng, L. Gao, Z.J. Huang, Y.Q. Wang and C.W. Chu, 1987, Plays. Rev. Lett. 58, 908. Yakhou, E, V Plakhty, A. Stratilatov, E Burlet, J.P. Lauriat, E. Elkaim, J.Y. Henry, M. Vlasov and M. Moshkin, 1996, Physica C 261, 315-322. "gang, S., 1992, Ph.D. Thesis (University of Illinois, Chicago, IL). You, Hoydoo, J.D. Axe, X.B. Kan, S. Hashimoto, S.C. Moss, J.Z. Liu, G.W. Crabtree and D.J. Lam, 1988, Phys. Rev. B 38, 9213. Zaanen, J., 2000, Nature 404, 714. Zaanen, J., and O. Gunnarsson, 1989, Phys. Rev. B 40, 7391.

Zaanen, J., A.T. Paxton, O. Jepsen and O.K. Andersen, 1988, Phys. Rev. Lett. 60, 2685. Zabinsky, ST, J.J. Rehr, A. Ankudinov, R.C. Albers and M.J. Eller, 1995, Phys. Rev. B 52, 2995. Zachariasen, W.H., 1931, Z. KristaUogr. 80, 137. Zandbergen, H.W., G. Van Tendeloo, T. Okabe and S. Amelinckx, 1987, Phys. Status Solidi 103, 45. Zech, D., H. Keller, K. Conder, E. Kaldis, E. Liarokapis, N. Poulakis and K.A. Mttller, 1994, Nature 371, 681. Zech, D., K. Conder, H. Keller, E. Kaldis, E. Liarokapis, N. Poulakis and K.A. Mfiller, 1995a, unpublished data. Zech, D., K. Conder, H. Keller, E. Kaldis, E. Liarokapis, N. Poulakis and K.A. Mfiller, 1995b, Total and site-selective oxygen isotope effect in YBa2Cu3Ox, in: Proc. Int. Workshop on Anharmonic Properties of High-To Cuprates, Bled, Slovenia, September 1994, eds D. Mihailovic, G. Ruani, E. Kaldis and K.A. Mfiller (World Scientific, Singapore) pp. 18-29. Zeiske, Th., R. Sonntag, D. Hohlwein, N.H. Andersen and T. Wolf, 1991, Nature 353, 542. Zeiske, Th., D. Hohlwein, R. Sonntag, E Kubanek and T. Wolf, 1992, Physica C 194, 1. Zhu, Y., and J. Taflo, 1997, Philos. Mag. B 75, 785.

Handbook on the Physics and Chemistry of Rare Earths VoL 31 edited by K.A. Gschneidner, Jr., L. Eyring and M.B. Maple © 2001 Elsevier Science B.V. All rights reserved

Chapter 196 FLUX

PINNING

Harald W. WEBER Atomic Institute of the Austrian Universities, Vienna, Austria

Comems 1. 2. 3. 4. 5. 6.

Introduction The mixed-state phase diagram Weak versus strong flux pinning Experimental techniques A flux-pinning anomaly: the "fishtail" effect Flux pinning by artificial pinning centres 6.1. Normal-conducting inclusions 6.2. Extended planar defects

187 188 192 195 204 213 213 216

6.3. Radiation-induced defects 7. Critical current densities and irreversibility lines 8. Field mapping and levitation forces 9. Summary Acknowledgements References

221 227 238 241 242 243

1. Introduction

Try to count the number of publications on flux pinning in ( R ) B a 2 C u 3 0 7 _ 6 (R-123), e.g. in the High-To Update (1987-2000) or in one of the more recently established databases (e.g., Superconductivity Database 1996), and you will realise that a comprehensive review of data accrued in these past fifteen years since the discovery of high-temperature superconductivity by Bednorz and Mtfller (1986) is just not feasible. Any solution of this problem will, therefore, necessarily contain a significant amount of personal views and preferences of the author as well of his (not unlimited) awareness of published work, and is certainly not aimed at a classification of research into better or lesser categories. The choices that had to be made were facilitated to a certain extent by the recently growing interest in the mixed-state properties in the Jc ~ 0 limit (cf. the Workshops on Flux Dynamics 1994-2000), i.e., the behaviour of the flux line lattice under (nearly) zero flux pinning conditions, and matters of flux lattice melting involving a first- or a second-order phase transition (e.g. Welp et al. 1996, Schilling et al. 1996, Roulin et al. 1996). These subjects are not considered below, except for a brief reference in sect. 2. Furthermore, an extensive review of flux pinning effects by "weak" collectively acting defects is available (Blatter et al. 1994), which does not need to be repeated, but will be referred to and compared with "strong" pinning characteristics in sect. 3. Work on ceramics, whose fluxpinning characteristics are strongly influenced by the details of the grain arrangement and, correspondingly, by the nature of the resulting weak link structures, sheds little light on 187

188

H.W. WEBER

the intrinsic flux-pinning properties and will, therefore, not be emphasised. As a result, the main thrust of this article will be placed on those flux-pinning effects which result in high critical current densities Jo even at high temperatures and in high magnetic fields; i.e., I will focus on those aspects of flux pinning which lead to applications of R-123 materials at or near the boiling temperature of liquid nitrogen. A further restriction that seems to be appropriate in view of the extensive literature, will refer to the chemical substitutions of the parent substance Y-123; i.e., we will mainly consider Nd substitutions, since the preparation processes and defect structures of this compound have been studied in most detail up to the present time. Hence, this article will be organised in the following way. After a brief review of the mixed-state properties and the nature of the flux-line lattice in sect. 2, I will address some characteristic features of weak and strong flux pinning depending on the nature of the defects responsible for pinning. Section 4 will be devoted to a discussion of experimental techniques commonly employed to assess the critical current densities in the various forms of material; i.e., single crystals, melt-textured monoliths or thin films and tapes. Section 5 presents an overview of a special feature in the magnetisation curve, which seems to be dominated by the "as-grown" defect structure in single-crystalline and melt-processed 123 materials, and which is commonly referred to as the "fishtail effect" because of a characteristic dip in the magnetisation curve both in increasing and in decreasing fields. It should, however, more correctly be described as an anomalous increase of Jo with magnetic field in between field ranges below and above the "anomaly", where Jc decreases with field - as expected from conventional flux-pinning considerations. Section 6 will then discuss the nature of intentionally introduced artificial defects, which are either produced by certain processing conditions (such as the addition of excess Y2BaCaO5 - Y-211 - in melt-textured materials) or develop "automatically" during the growth process (as in R-123 thin films) or have to be introduced artificially by suitable types of radiation. The results of these procedures and the corresponding flux-pinning action of these various defect structures will be presented in the form of selected examples in sects. 7 and 8, where the resulting critical current densities, the irreversibility lines, the renmant magnetisation as well as the levitation forces will be presented. I will conclude with a brief summary in sect. 9.

2. The mixed-state phase diagram The basic features of the mixed-state phase diagram have been established many years ago by Abrikosov (1957) on the basis of Ginzburg-Landau (1950) theory. It strictly refers to thermodynamic equilibrimn conditions and to a "perfect"; i.e., totally homogeneous and defect-free, superconducting material. As shown in fig. 1, three phases can be established depending on the magnitude of the magnetic field: the Meissner state (Meissner and Ochsenfeld 1933), which is characterised by complete flux expulsion and is stable below the lower critical magnetic field Hal; the normal conducting state, in which the magnetic field penetrates the sample completely and homogeneously above the upper critical

FLUX PINNING

T

(K)

189

To

Fig. 1. Mixed state phase diagram of conventionaltype-IIsuperconductors.

Fig. 2. Three-dimensionalview of the flux line lattice in niobium, as determined from neutron diffraction(Weberet al. 1973).

field Hc2; and, in between, the mixed state, which is characterised by the coexistence of superconductivity and some magnetic-field containing regions. The clarification of the microscopic nature of this state was certainly one of the most fascinating tasks in superconductivity since the theoretical prediction by Abrikosov (1957) that it consisted of a quantized array of magnetic flux quanta (q}0= h/2e) forming a regular two-dimensional lattice of straight "flux lines", arranged parallel to the magnetic field direction and each surrounded by a set of supercurrents circulating around the centres of the flux line cores in closed loops ("vortices"). The experimental verification of this prediction took almost 10 years, and was achieved by neutron diffraction (Cribier et al. 1964, Schelten et al. 1971, Weber et al. 1973) and by the "decoration technique" (Tr/iuble and El3mann 1966). A three-dimensional view of such an undisturbed flux-line lattice at low magnetic fields is shown in fig. 2 (Weber et al. 1973); a comprehensive review of the subject was prepared by Brandt (1995). Since all of the high-temperature superconductors (HTS) belong to the same class of superconductivity - they are type-II superconductors - not too many changes in the mixed-state phase diagram were expected initially. However, this turned out to be an oversimplification and, in fact, an erroneous assumption. The number of phase diagrams that has appeared in the literature during the past few years is uncountable; an example pertaining to Y-123 is shown in fig. 3 (Blatter et al. 1994). However, before going into the details of this figure, a few words of caution seem to be appropriate. As pointed out in the last paragraph, the mixed-state phase diagram of fig. 1 strictly refers to the thermodynamic equilibrium state and is, by definition, only valid for a superconductor in which the flux lines are subjected only to the repulsive interactions between them, but flux-pinning interactions between defects and the flux-line lattice, which would render the free energy of the system position dependent, do not exist. This condition cannot be met by high-temperature superconducting materials'up to the present, since structural defects are always present, although their pinning forces may become extremely small or negligible, at least at high temperatures. The pinning force is strongly dependent on the material under consideration, i.e. the apparent reversible regime in the phase

190

H.W. W E B E R

(a)

~[h fvortex liquid 1 ~[Jll[l~ ', fluctuation /

~H

0 B .UoH~I(0) 10-3 0

T

T~

Fig. 3. Mixed-state phase diagram of high-temperature superconductors (Blatter et al. 1994): (a) phase diagram with parameters chosen to represent Y-123; (b) enlarged view of the low-field melting line.

diagram may range from ~0.4 ~
FLUX

,

PINNING

191

(b)

(a) :~y

pancake vortex

X

J 0

I

I

• y

losephsonstring

Fig. 4. Pancakes in high-temperature superconductors: (a) stack of aligned pancakes for H J[c (Clem 1991); (b) decoupled pancakes in a strongly two-dimensional HTS (Blatter et al. 1994).

An additional phase boundary line must be added at relatively low fields, which refers to the nature of the flux line itself and which is again of special importance for the low-dimensional materials, although some indications for its relevance also in Y-123 (predominantly at low temperatures, where the c-axis coherence length is very small) were deduced by Schalk et al. (1992a, 1994) from measurements of the angular dependence of Jc in Y-123 thin films and the corresponding analysis in terms of the pinning potential. In essence, the layered character of the superconductors leads to a break-up of the straight (continuously connected) flux line into a series of layers carrying the supercurrents alternated by "blocking layers", which may either be nonsuperconducting (in the extreme two-dimensional case, Lawrence and Doniach 1971) or only weakly superconducting. These superconducting layers form stacks of "pancakes" (Pearl 1964, Clem 1991, Bulaevskii et al. 1992), which are coupled by magnetic interactions and by Josephson strings, thus aligning them along the field direction much in the same way as a regular flux line (cf. fig. 4a). However, as the field increases, the correlation between the pancakes becomes weaker and they finally decouple (fig. 4b) at a certain field HDC, which could be directly observed in neutron diffraction experiments on Bi-2212 single crystals (Cubitt et al. 1993). This "decoupling" line, which is fairly independent of temperature, finally joins the melting line at elevated temperatures and certainly represents a thermodynamic equilibrium phase boundary that is related to the nature of the magnetic microstructure. Obviously, the question of whether or not a continuously connected flux line or a set of independent pancakes determines the magnetic microstructure of the mixed state, will play a significant role for flux pinning. Apart from details of the flux-pinning interaction (cf. sect. 3), one important issue was pointed out early on by Kes et al. (1990), i.e., that the response of decoupled pancakes will always be determined entirely by the c-axis component of the magnetic field. The discussion so far has been tacitly based on the assumption that the magnetic field is directed parallel to the crystallographic c-axis of the superconductor. If we drop

192

H.W. WEBER

this assumption, the general shape of the phase diagram is modified only moderately. The corresponding upper critical fields He2 will be much higher, the lower critical fields Hcl lower by the same facto1, although not too much is known, mainly because of experimental problems which would require sample alignments of better than 1° in most cases. However, the magnetic microstrucmre is changed and this is again more pronounced in the case of two-dimensional materials. The first striking effect, occurring also in nearly three-dimensional materials, was predicted by Tachiki and Takahashi (1989) early on and is related to the formation of the flux-line lattice. Because of the "normalconducting" flux-line core, the lattice will gain energy if the flux lines are located on weaker superconducting or, even better, normal conducting regions. These are present in all HTS in a more or less pronounced form due to the existence of the "blocking layers", which are characterised by a (slight) suppression of the order parameter. Hence, the flux lines will automatically move to these regions, i.e., into the space between the fully superconducting CuO2 planes ("a,b"-planes), thus preventing the formation of a regular long-range ordered flux-line lattice of hexagonal or square symmetry. Instead, they will form chains in between the CuO2 planes, which were indeed observed by decoration experiments (Dolan et al. 1989, Gammel et al. 1992). This is called the "intrinsic pinning" effect and falls into the category of thermodynamic equilibrium phenomena. The second effect only plays a role in truly two-dimensional materials, where the blocking layer is sufficiently thick and prevents the formation of a finite order parameter. In this case, the circulating supercurrents forming the flux lines are completely suppressed and replaced by Josephson currents with different dimensions and physical properties (Clem and Coffey 1990). Obviously, this change of the magnetic microstructure ("Josephson vortex") will also have a significant impact on the flux pinning properties in two-dimensional HTS materials.

3. Weak versus strong flux pinning

The basics of flux pinning were comparatively simple in the metallic low-temperature superconductors, since "strong pinning" prevailed in most cases, i.e., the pinning potentials were deep and matters of thermal activation were not too important because of the low temperatures involved (cf., e.g., Campbell and Evetts 1972, Ullmaier 1975, Weber and Hittmair 1979). It was, therefore, possible to model the elementary fluxpinning interaction between a certain defect and a single flux line fairly successfully, but the summation of these individual forces (Pro) to obtain the bulk volume pinning force P v = - J o × B, and hence the critical current density, or - more correctly - vice versa, turned out to be tedious (e.g., Labusch 1969, Dew-Hughes 1971, 1974). Most of this work on "model systems", i.e., homogeneous superconductors with just one type of deliberately introduced defects, indicated that the summation was in between the "direct" summation law, where the individual elementary pinning forces Pm were just multiplied by the defect density and added up, and the "statistical" summation, which predicted a quadratic dependence of the bulk pinning force on Pro. Also, at that time, Larkin and

FLUX PINNING

193

Ovchinnikov (1974, 1979) proposed the "collective" pinning theory, which introduced certain collective actions of several defects over a certain correlation length or volume for the first time, a theory that represents the basis for most flux-pinning considerations in high-temperature superconductors because of their generally "weak" pinning potentials. In order to put these general remarks onto a more quantitative basis, three characteristic parameters and their actual magnitudes in low- versus high-temperature superconductors have to be considered. First, the critical current density Jc can never exceed a certain maximum value, which is defined by the material-dependent "depairing" current; i.e., the maximum loss-free current Jc,max the superconductor can sustain before breaking up the Cooper pair correlation. Since Jc,m~× e( Hc/)~, i.e., proportional to the thermodynamic critical field divided by the magnetic field penetration depth X, the intrinsic material dependence is obvious and provides at the same time a natural distinction between lowand high-temperature superconductors, since Hc is relatively large (in particular compared to the upper critical field) and 3. is relatively small in the former case, whereas the opposite is true in the latter. Thus, Jc/J~,max will be of the order of 10 1 for conventional superconductors versus <10 -3 in HTS. Second, we are concerned with high operating temperatures in the case of HTS, which render thermal fluctuations far more important and lead to thermally activated depinning or melting. This effect is conveniently characterised by the Ginzburg number Gi (e.g., Blatter et al. 1994), which is proportional to the square of T~ divided by the square of the minimum condensation energy ((xHZ~3e, where ~ is the coherence length and e is the mass anisotropy parameter, (ma,b/m~)l/2). Numbers for Gi dramatically emphasise the problem: Gi ~ 10 8 in low-temperature superconductors, but Gi ~ 10 2 in HTS. Third, even quantum fluctuations affect the phase diagram (cf. sect. 2) and can add to the creep behaviour of the vortices. This effect differs in low- and high-temperature superconductors by a factor on the order of 100. Finally, since all of the above three effects are strongly affected by anisotropy (either through ,~ or directly through the anisotropy parameter e), all flux-pinning effects in the HTS will depend on the directions of magnetic field and current flow with respect to the crystallographic axes in the material. The difference between strong pinning and weak collective pinning can be visualised as follows (fig. 5). In the first case, a defect with a diameter of at least 2~(T) is required to provide the flux line with the highest possible energy gain by saving the condensation energy of the ("normal conducting") flux line core (/~0H2/2 per unit volume). In the best case, the defect should, therefore, be normal conducting itself, but superconducting inclusions with significantly different thermodynamic properties of their superconducting state (compared to the surrounding matrix) will also serve the purpose. The latter is often referred to as ATc, ~ / c 2 , or Ag-pinning, where g is the electron mean free path, which primarily affects the Ginzburg-Landau parameter ~c and the upper critical field Ho2 (Campbell and Evetts 1972, Freyhardt 1971 a, Antesberger and Ullmaier 1974, Adaktylos and Weber 1977). Depending on the size and the shape of the defect, the energy gain of the flux line reaches its maximum for normal-conducting inclusions and is given by the condensation energy multiplied by the length of the defect along the field direction and by the cross-sectional area of the flux-line core or of the defect in the plane perpendicular

194

H.W. WEBER

lll ) z '.

~ ~

...'

(c)

Lc

Fig. 5. Flux pinning in superconductors: (a) strong pinning through the core interaction by normal conducting precipitates of various shapes; (b) schematic view of the pinning potentials without (top panel) and under the action of a Lorentz force (bottom panel); (c) schematic view of pinning of a single flux line by the collective action of many small point pinning centres (after Ullmaier 1975).

to the field. This energy represents an upper limit of the pinning potential for H ---+0, which will, of course, in general decrease with temperature and magnetic field. At fixed temperature, the Lorentz force PL acting on the flux lines will shift their spatial distribution and result in a "wash-board"-like arrangement of pinning potentials (fig. 5b) leading eventually to depinning for PL = - P v - Since this "critical state" equation does not refer to a thermodynamic equilibrium situation, depinning is always assisted by any energy transfer to the system, in particular thermal energy, as was pointed out by Anderson (1962) early on ("flux creep"). Contrary to the above, weak pinning prevails if the defect structure consists of a large number of very small defects ("point" pinning centres), the spatial distribution of which m u s t fluctuate over a certain volume. In this case (fig. 5c), the entire ensemble of these defects interacts collectively with the flux line (we consider again the limit H ---+0) and exerts a pinning force over distances on the order of ~ in the plane perpendicular to the field and of the "correlation length" Lc along the field direction, which is largely determined by the elastic properties of the flux line. The actual scales of fig. 5c are more appropriate for the low-temperature superconductors, because ff is rather large compared to the defect sizes, but the general picture remains the same in high-temperature superconductors, if the defect size does not exceed 0.5-1 nm, i.e., is clearly smaller than ~. I f we now increase the magnetic field, the usual interactions between the flux lines will start to play a role and lead to a change in the "correlation volume"; i.e., the coherence length ~ in the above flux-pinning considerations has to be replaced by the "transverse correlation length", Rc. As a consequence, several flux lines can be pinned simultaneously by the same ensemble of defects (collective pinning of three-dimensional flux-line bundles).

FLUX PINNING

195

Under the influence of thermal activation, a rich variety of scenarios can emerge, which has been described by Blatter et al. (1994) in much detail. In general, the critical current density will decrease more rapidly with temperature and the character of the flux line lattice can change into a "glassy state". In addition, the dynamics of the system change and the pinning potentials depend on current density in a non-linear way (in contrast to the original flux creep model by Anderson 1962). I f we consider the lowerdimensional high-temperature superconductors, where the continuous flux line breaks up into pancakes, as discussed in sect. 2, individual single-pancake pinning, individual pinning of stacks of pancakes, collective pinning of two-dimensional bundles and/or collective pinning of three-dimensional bundles can take place. Furthermore, transitions between these possibilities may occur depending on the size of the coherence length ~(T), which grows with rising temperature and may, therefore, lead to stronger coupling across the blocking layers (H II c) and, hence, to a more three-dimensional behaviour at elevated temperatures. In all cases, (collective) creep or even thermally assisted flux flow (Kes et al. 1989) prevails and affects the data obtained from experiment, an issue that will be addressed in the following section. In conclusion, as pointed out in the introduction, the results to be presented in sects. 5-8 are not aimed at a detailed discussion of these fundamental aspects of flux pinning and flux dynamics, but rather at possibilities to improve Jc in the 123 superconductors. We must be aware of the fact, however, that the "Jc's" quoted in the literature often do not refer to the "true" critical current densities, since the data are affected by creep and relaxation, and should rather be quoted as shielding current densities Js. Furthermore, we will refrain from judgements of the nature of the boundary line, where Js goes to zero, and will refer to it as the irreversibility line. This characteristic parameter is subject to the same restrictions as mentioned above; i.e., sensitive to the resolution of the experiment and its time scale. It is useful to note that the shape of this curve (H IIc) follows a power law (with an exponent of 1.5) for more three-dimensional HTS, as predicted by Malozemoff et al. (1988) on the basis of a depinning argument, but an exponential law for two-dimensional systems, as shown schematically in fig. 6.

4. Experimental techniques Jc and the irreversibility line can be measured with several techniques, but we are usually not completely free to choose among them, since the form of the material under investigation imposes certain constraints, which can only be met by one or several special techniques. In principle, Jc is always determined from the critical state equation

OH

Pc = Jc × B = ( r o t H × B) = (rotB × B)-a~ = -Pv, 0/5

(1)

where OH/OB can be set equal to 1 in HTS. Hence, the critical current density can be assessed from a direct transport critical current measurement or, magnetically, from the

'/'

196

10

H.W. WEBER

'\i'

"3D"

"2D" 8

m o 7-

6

Bi-2212 TI-2223

\

=f ._~ 4 7o

\

2 0

0,0

Y-123 T1-1223

\ HI[C ,--~--v---~ , , ~ , i . ~ , 0,1

0,2

0,3

0,4

0,5

T/To

0,6

0,7

\

0,8

\

\

\\ 0,9

1,0

Fig. 6. Schematic representation of the irreversibility lines in HTS. The "3D" case corresponds to a power-law dependence of Hi,. r on temperature, the "2D" case to an exponential dependence.

corresponding flux-density gradient built up by the pinning centres. The latter is usually not directly accessible and must be evaluated from models, in most cases from the Bean model (Bean 1962). Direct transport measurements are generally employed on thin films, tapes and coated conductors. They quickly reach their natural limits as the currents go up and/or the crosssection of the conductor increases. They are fast, i.e., can be started immediately after the magnetic field is set, and trace the I - V characteristics, which is of course most valuable if flux creep or thermally assisted flux flow (Kes et al. 1989) prevails, because the electric field is directly determined. On the other hand, Jc must be defined by a criterion; e.g., 1 gV/cm. It will be affected by the presence of stabilising materials, the silver tube or conducting substrates, and thus generally refers to the "overall" Jc, which can be converted to the critical current density of the superconductor if the shape and the volume fraction of the superconducting material are known. Magnetic methods are required for single crystals (because of their generally very small size, the corresponding contact problems and the huge critical currents) and for melttextured bulks (because of their enormous current requirements). Several possibilities are available: dc and ac techniques, and among the dc methods "global" and more "local" techniques. Starting with the global dc techniques, Superconducting Quantum Interference Devices (SQUID's) have been used most extensively, followed by Vibrating Sample Magnetometers (VSM's). In both cases, the magnetic moment of the sample is measured as a function of magnetic field, but the "timing" of the experiment can vary by orders of magnitude. Taking, e.g., the custom-built high-field (8 T) SQUID operated by our group, the settling time of the instrument after applying a certain magnetic field amounts to about 10 minutes followed by the measuring cycle (where the sample is slowly moved up and down in a set of pick-up coils), which takes at least another 5 min. Therefore, if strong creep prevails, we obtain information on the magnetic moments after a relaxation period of 10 min and then averaged over an additional period of 5 min (Kritscha

FLUX PINN1NG

197

et al. 1990a). On the contrary, the VSM can collect data in continuously (but slowly) changing external fields and, therefore, provides data in a different "time window" of the relaxation process. In both cases, the moment m is then converted into the magnetisation, M=m/V, where V is the sample volume, plotted as a function of the external field and used to calculate the critical current densities via the Bean model and/or its extensions for finite geometry and demagnetisation effects (Gyorgy et al. 1989, Sauerzopf et al. 1990, Wiesinger et al. 1992a, cf. also Brandt 1995, 1996, and references therein). The third global technique, to be discussed briefly, is torque magnetometry, and I take the example of a capacitance torquemeter (Griessen et al. 1977). Here, the torque • exerted on a superconducting sample with a magnetic moment m in an external magnetic field H , is given by ~ = m x #0He = m/~t0Hesin a, where a = a(m, ~toHe) is the angle between the magnetic moment and the external field. Obviously, the torque disappears if the moment is parallel to the field direction. A certain disadvantage is the necessity of applying the field out of the main crystallographic directions (e.g., at an angle of 5 ° with respect to the c-axis) and the requirement of re-calibration at each temperature. All further procedures are the same as above, i.e., the moment is extracted from the data, converted into the magnetisation and evaluated in terms of certain models to obtain Jc. The main advantage of measuring the torsion with a capacitance lies in its very high sensitivity and, in particular, in its very fast response. Hence, magnetisation relaxation can be studied at a very early stage and, if combined with a fast sweeping magnet, be extended to the so-called dynamic relaxation regime, a concept that was originally introduced by Pust et al. (1990) and explored in much detail by the group at the Vrije Universiteit (VU) Amsterdam (Quaford et al. 1992, Lensink 1993, Schnack 1995, van Dalen 1995). Of course, an important question should be asked immediately; i.e., what do we really extract from such magnetic measurements in terms of the critical current density? This matter was addressed by Werner et al. (Werner et al. 1996, Werner 1997), who started to compare experimental results on the same single crystal, but obtained from different techniques. An example comparing torque and SQUID magnetometry on an Yb-123 single crystal (H 11c) is shown in fig. 7, where the shielding current density J~ is plotted versus the induction B in the crystal (Werner et al. 1996), as evaluated form a modified Bean model (Wiesinger et al. 1992a). Details of the shape of the Js-B curve are not of interest right now, but will be considered in sect. 5. The uppermost five curves show the dynamic relaxation response obtained by torque magnetometry under different field sweep rates. The curves marked by the symbols refer to conventional relaxation after 10 and 15 minutes, respectively, starting at several fields from the uppermost curve (40 mT/s), and the bottom solid line shows the data taken by the 8 T SQUID under the conditions mentioned above. Apart from the very nice agreement between the conventional and the dynamic relaxation rates (Werner et al. 1996), the data obtained from the torque and the SQUID at comparable relaxation times are very close and demonstrate the feasibility of such magnetic measurements and of the' evaluation methods for obtaining reliable data on the current density J~. Turning next to more "local" techniques, the magneto-optical Faraday effect will be discussed first (Alers 1957, Kirchner 1968, Huebener et al. 1972, Koblischka and

198

H.W. W E B E R 5.5

i

5.0

I

Goetl 50K

I

I

i

i

i

I

~

40 mT/s

~ 7 2 7 " , . ' . ~

4.5

/4~

4.0

~

. j ~ A

.......... 30 mT/s 20 mT/s

¢~,~.

.....

~.'~.~2~,~..

. . . . . . . 10 mT/s

N-"

1.0 0.5

z~-- J (40 mT/s) after 10 min ~ J (40 roT/s) after 15 min

0.0

,

-1

I

0

,

I

1

i

I

2

,

I

3

i

I

4

i

I

5

i

I

6

,

I

7

i

8

a (T) Fig. 7. Comparisonof experimentaldata obtained by SQUIDand by torque magnetometryon the same single crystal (Werner 1997).

Wijngaarden 1995, Wijngaarden et al. 1996). Here, the penetration of the flux front into, e.g., a single crystal is traced by the rotation of the polarisation vector of linearly polarised light when passing through a certain "magneto-optical" layer (e.g., a thin film of EuSe). The parts of the layer which are subjected to a magnetic field will provide us with a different image compared to the field-free regions (dark versus bright, depending on the set-up of the polarisation analyser) with a spatial resolution on the order of 0.5 ~tm. The magnetic resolution depends on the magneto-optical layer used for the experiment, but is generally on the order of 5 mT, and the time resolution is immediate. Two important conclusions can be drawn from such an experiment. First, the visual inspection of the flux penetration process leads to a crucial check of the models employed for the above evaluations of the magnetisation curves in terms Of Jc. As shown in fig. 8, all the general features of the Bean model in its modified form for a finite geometry are beautifully confirmed by experiment along the entire magnetisation loop from 0 to +7 T to zero and then to - 2 T. Thus, evaluations of Jo as described above are feasible and lead to the right answers. Second, the details of the flux patterns, in particular the local distributions of the field Hz at the sample surface, can be converted accurately into the current patterns (Wijngaarden et al. 1996) and thus provide us with more local information on the critical current density than the global techniques described above. The latter also holds for the last dc technique I will discuss in the following, i.e., the application of multi-Hall-probe arrays for the direct recording of the flux density gradients along the radius of a superconductor. It was invented about 25 years ago (Weber et al. 1976, Adaktylos et al. 1977), forgotten, and re-invented (Majer et al. 1996), although with significantly improved technology. In this case, an array

FLUX PINNING

199

.... ~ ¸,¸4¸

0.1 T

0.2 T

0.4 T

1.5 T

3T

7T

3T

1.5T

0T

-0.4T

-0.ST

-2T

Fig. 8. Magnetisation loop (0-~ 7 0 - * 2 T) recorded by the magnetooptical Faraday effect on a single crystal (Werner 1997).

of approximately 10-15 very small Hall probes (e.g., 10×10~tm 2) are manufactured from GaAs/A1GaAs heterostructures and arranged at equal spacing of, say, 100~m. The Hall voltages, therefore, directly display the flux profile, i.e., the z-component of the local induction at the sample surface. This can again be converted into critical current densities with similar algorithms as in the case of the magneto-optical Faraday effect. The advantages of this technique are manifold. First, the spatial resolution can be adjusted to the experimental requirements (active areas of 1 × 1 ~m 2 have been achieved and the spacing can be made according to sample size and/or the type of experimental investigation envisaged). Second, the time resolution is immediate, and the system can also be operated under ac conditions. Third, magnetisation loops or relaxation measurements can be made at any of the available positions "within" the sample. Of course, the amount of data handling is quite extensive, in particular since the individual Hall probes may require individual calibrations (deviations of their field response from linearity) and re-calibrations at different temperatures. Many novel experiments based on this technique have been reported in the literature (e.g., Abulafia et al. 1995, 1996, Yeshurun et al. 1996, Doyle et al. 1997). For the purpose of this section, I wish to concentrate on two results, which are shown in fig. 9 (Werner 1997). First, the flux profile (fig. 9a) obtained at a certain external field (~280mT, 60 K) again shows all the characteristic Bean-like features and a (smeared-out) increase of the field at the sample edge as expected from demagnetisation. Second, the comparison between the local Hall-probe measurement and SQUID magnetometry (fig. 9b) is as favourable as in the case of torque magnetometry, i.e., the current densities J~ deduced

200

H.W. WEBER 0.30

i

#1

i

#2

i

#3

i

#5

#4

i

#8

0.28

i

#7

#8

#1o

#11

M3944 60 K

- e - p o l l = 0.28 T

\

i

#9

0.28

\\\\\\\ crystal

0.24

i,

\

rn ~

O\

edge of sample

022

#tl

i center of sample '\\\\

~ Hall array

0.20

:

\,

probes

(a) I

0.18 -600

-500

-400

i

-300

I

I

I

I

-2oo

400

o

lOO

sensor position (gin)

!

~

i , i 0 0 ~ O 4 ' O O 0

9

8

M3944 60 K unirradialed

II

4u s

E

~

i

~

~

i

• -

v •

,

\t

!

i

o

~'

/

3

i

~

.... -.., /

1

,

.

"4

Hall

loop

SQUID

, l .

(b) I

0.0

i

I

02

i

I

I

I

I

I

0.4

0.6

08

1.0

12

i

I

1.4

i

16

B (T)

Fig. 9. (a) Measurement of the flux profile in a single crystal with a multi-Hall-probe array: B~ versus sensor position (Werner 1997). (b) Comparison of experimental data obtained by SQUID and Hall probe magnetometry on the same single crystal (Werner 1997).

from both kinds of experiment agree very well and, therefore, confirm the basic concepts of these techniques. The last technique to be discussed in the context of Jc measurements is an ac method proposed by Campbell (1969) and modified by Rollins et al. (1974), which is particularly well suited for measurements on bulk samples; e.g., melt-textured HTS. A schematic view of its basic principles is shown in fig. 10. Assume that a Bean-like flux profile has been established by a certain dc field well above the full penetration field. If a small ac ripple field of amplitude h~c is superimposed parallel to the dc field, the flux profile near the sample surface is modified accordingly (as shown by the shaded area in fig. 10). With

FLUX PINN1NG

201

Fig. 10. Schematicrepresentationof the flux frontnear the sample surface upon superimposing an ac ripple field of amplitude h~c onto a large dc field H~c (Frischherz et al. 1995). increasing hac the "modified" flux profile moves further into the interior of the sample and eventually affects the entire flux-density gradient. Hence, the idea of such a "flux profile" experiment consists of measuring the induced voltages in a set of compensated pick-up coils (one containing the sample) as a function of ac amplitude at constant dc field and temperature. The in-phase signal Sin of the lock-in amplifier in the superconducting state is given by Sin - a'(a + b) h2a + 13hac + Y = ah2c + [3hac + y, jo

c

(2)

where a',/3, and • are instrumental constants and a, b are the geometrical dimensions of a rectangular sample in the plane perpendicular to the field (c >> a,b; H I]c). Sin(ha~) is then fitted to a parabola to obtain a and the critical current density is calculated from

Sn a + b J c - hac a a b '

(3)

where Sn/hac is the slope of the signal in the normal conducting state. In practice, both the real and the imaginary part of the signal are measured at the same time, the frequency is kept relatively low (9 Hz), and the amplitude is varied between 0 and 10-20 mT. The latter can be avoided, if the version introduced by Rollins et al. (1974) is employed, where signal averaging and phase-sensitive signal analysis at a fixed ac field amplitude are sufficient. In principle, this method is very well suited to determine critical current densities in large samples, unless the flux profiles become too steep to be noticeably changed by the ac field over a significant penetration range into the interior of the sample (such as for instance in melt-textured samples at 4.2 K). A comparison with SQUID magnetometry (Frischherz et al. 1995) leads to similar results as shown in figs. 7 and 9 for torque and Hall probe arrays; i.e., Jc as deduced from this ac technique is larger (because of its fast response) than that obtained from a SQUID experiment (by a factor of up to 2, depending on relaxation).

202

H.W. WEBER

To conclude this section, a few words on the experimental assessment of the irreversibility lines seem to be appropriate. As pointed out in much detail above, the timing ("time window") of the experiment represents a decisive factor for the apparent magnitude of Jc. Since we are now looking for methods to determine the exact location in the H, T plane, where Jo becomes zero, the timing will again play a crucial role. In addition, the criterion Jc = 0 is extremely ill defined, since it entirely depends on the resolution of the experimental technique employed. I will choose two examples (SQUID magnetometry and ac susceptibility) to demonstrate the salient features of the problem. In the dc case, there are two ways of determining the irreversibility line: one is to sweep the magnetic field at constant temperature and take the point where the loop closes, i.e., the hysteretic behaviour disappears (fig. 1 l a), the other is to sweep the temperature at constant field and take the point where the zero-field-cooled (ZFC) and the field-cooled (FC) branches merge (fig. 1 lb). The latter is more frequently used, because it is generally easier and faster to vary the temperature in small increments. Obviously the results will be quite arbitrary, since the curves always merge tangentially and will strongly depend on the resolution of the set-up and the time scale chosen. Distinct improvements can be achieved by noting that the measuring principle involves a movement of the sample between the set of pick-up coils in a slightly inhomogeneous field, which can be visualised to be equivalent to the superposition of a small ac field and which results in a distortion (fig. 12) of the SQUID signal response curves, if dissipative processes ("hysteresis losses") set in (Kritscha et al. 1990a, Suenaga et al. 1992, Sauerzopf et al. 1992). In this way, the "arc =0"-criterion can be approximated by 104-105 A m -2 (depending on sample size). If this more sensitive criterion is used, all differences, mentioned occasionally in the literature, between the "loop" and the ZFC-FC experiments disappear, thus confirming equilibrium conditions in both cases (Kritscha et al. 1991). Even more commonly used are ac techniques, which are based on the detection of energy losses of the superconductor, much in the same way as discussed above for the signal distortion in the SQUID (for reviews, cf. Clem 1992, Frischherz et al. 1995, G6m6ry 1997). A wealth of information on the frequency and amplitude dependence of Z ~ and Z ~ has been accrued for all kinds of HTS and for magnetic field orientations parallel and perpendicular to the crystallographic c-axis. The main points are the following. If the superconductor is in the critical state, a peak in the out-of-phase signal Z" occurs, when the flux front reaches the centre of the sample. Of course, this peak indicates the presence of a certain critical current density Jo (which is given in the simplest case of an infinite cylinder by hac/r) and, therefore, defines a certain sampledependent Jc criterion (of similar order as in the above dc case, i.e., 105 Am-2). On the other hand, if strong flux creep prevails, e.g., thermally assisted flux flow, X" will also show a peak (Geshkenbein et al. 1991) at a certain flux flow resistivity PXAF~,which is given by pTAFv(H,T)= ~t0~vr 2 (v is the excitation frequency, r the sample radius). Since in most cases of interest, both effects will contribute to the experiment at the same time, we expect a certain amplitude dependence (from the critical state equation) and a certain frequency dependence from flux creep or flow. Although the analysis is obviously not trivial, the peak of Z" is still often used to characterise the irreversibility point because

FLUX PINNING

I

'

I

'

I

I

'

10

203

I

'

I

I

I

soo | ~ . . . . . . . .

isothermal

1 1

[ \'-,

magntizatio. [I / \~

T = 77 K

6~X

c,,-1

-5oo! S . . . . . . . .

. . . . .

0

2

4

/

6

8

v

E

I

I

[

3.0

r

3.5

I

i

4.0

I

I

I

I

I

4.5

5.0

5.5

6.0

6.5

i

~o H (T) I

I

I

I

I

. . . . .

. . . . -2

oS/

c" -i

\

Tit r = 78.5 K

-4 v

ZFC-FC magnetization

E

o

goH = 4 T

-6 I

76

i

[

78

~

I

80

i

f

82

i

I

84

T (K) Fig. 11. "Standard" methods to determine the in'eversibility line from dc measurements on single crystals (Frischherz et al. 1995): (a) Isothermal magnetisation ("closing" of the loop); zero-field-cooled/field-cooled magnetisation ("opening" of the branches).

of the relatively high experimental accuracy with its determination (fig. 13). Alternatives are extrapolations of the loss signal to zero at high temperatures (Yacobi et al. 1992), an extrapolation of the in-phase signal X~ to zero at high temperatures, or the analysis of higher harmonics, where again the peak or extrapolations can be employed. A comparison of these various evaluation possibilities is shown in fig. 13 (Wacenovsky 1993) for a melt-textured Y-123 sample. Taking the example of 1 T, the peaks of Z" and of the third harmonic lead to the same irreversibility temperature Tirr, and the same holds for the extrapolations of ) ~ and of the third harmonic, the difference being -0.5 K. A more

204

H.W. WEBER |

-0.8

_01 0

I DL14

t

~?"\

>:~;'

)~t"..)"'.\\ ~ / -0.7

14

16

T

K

18

'~~reversible x,,,,> \ ~

.

;~,~.~ ............................. % \ ""'"="""" __~-:~%5 . . . . . . .

-0.9

.

/i~'/

20

.

.

~"

....

~..y/

~.;'2-2~-S-&vz;~L~'~* :~='~

"-_~.-"

irreversible

-1.1

'

30

'

40

'

"

50

Time (s)

'

~ T

15.97 K 16"34 K

"""'"- T = ~- - T = ......... T = -----T= __ - T = - T=

"

60

17.10 K 17.51 K 17.90 K [8.31K 18.71K 19.13 K

'

70

Fig. 12. Distortion of the SQUID output signals by hysteresis effects in a single crystal (Kritscha 1991).

complete comparison of data obtained at different frequencies and field amplitudes (peak in X", extrapolation of X" to zero: "onset") as well as by SQUID magnetometry is shown in fig. 14, again for a melt-textured Y-123 sample (Frischherz et al. 1995). Systematic dependencies are found in all cases, including the SQUID measurement, the "frequency" in the latter case being determined from the speed of the sample movement between the pick-up coils. In summary, if the frequencies and the ac amplitudes are kept low, the differences between the various experimental parameters lead to discrepancies in Tirr on the order of 1 K. In conclusion, sufficient control of the experimental parameters and a critical assessment of the characteristics of each techniques will in general lead to reliable data both on the current densities and the irreversibility lines. Direct comparisons of published data are still difficult, in particular when the experimental parameters are not explicitly specified.

5. A flux-pinning anomaly: the "fishtail" effect It may appear strange to start the discussion of flux-pinning results with an "anomaly", but almost every 123 single crystal, many other high-Tc single crystals (such as, e.g., T1-1223) and several melt-textured materials (e.g., Nd-123) show the so-called "fishtail" effect, which was given this name because of a characteristic dip in the magnetisation curve (fig. 15a, D/iumling et al. 1990). The development of the "second peak" is obviously related to an anomalous increase of the critical current density with magnetic field. However, it turned out soon (fig. 15b, upper panel) that such a dip is not a necessary prerequisite of the Jc anomaly, since it can be masked in a magnetisation curve (Kritscha

FLUX PINNING 0.2

MPMG-3, HIIc

o.o

- 0.1 m T -- ~ 0.5 m T ..... 1.0 m T

205

normiert

in-Phase-Signal,

i

)/oT

-0.2 ,/

-0.4 o9 ~:~ - 0 . 6 o

;s

"I It

/I ,'l ,'l ,,I ,'1

-o.8

;I /I ,q /I ,,I

//

-1.o

7'0 . . . .

7;

i

i

i

,

.'o . . . .

,

~

i

t

*

I

90

T(K) 0.5

MPMG-3, HIIc

o.1 mT

--.....

normierL

out-of-Phase-Signal,

Q.5 r a T t,0 mT

0.4 0~

8T

3T

IT

0.3

~ 0.2

,,I // / I / /

0 0.1

,'/ ;I ]I / / ,'1 ,,,~

// /J

..j,

[ I If

.:~',

.,

0.0 ~

,

,

, 715

~

~

~

~

7

•

80

i

~

i

i

I

85

,i

J

I

~

I 90

T(K) i

3

~','"

i

,

i

MPMG-3,

r

i

,

i

,

T

i

i

i

i

i

Hlic

3. O b e r w e l l e ,

normiert

~

i

i

*

..... ---

1.0mT 0.5rot

- -

0.lint

o9

L

1T

f.T..]

8T

"~

"" "-'~

I

70

i

i

i

i

I

75

,' ~d

, / " 't

i

i

i

i

I

80

i

i

/ i

i

i

"'

i

,

I

85

i

i

i

i

I

90

T(K) Fig. 13. Measurement of the ac susceptibility on a melt-textured superconductor (Wacenovsky 1993). Top panel: in-phase signal; middle panel: out-of-phase signal; bottom panel: third harmonic.

206

H.W. WEBER 91

, , ,,-,,i

,

90

* J ° J 1 onset 1

•

:11

H//ab ,,g v

SQUID

/

I-,8

ie~._~E

H//c . ~ n peakjj~ ~ "

m~ m

[] 87

i i ,,,.d

0.01

one'set

.

,,,,,,,I

0.1

.

,,.,,,d

1

,

10

, ,l,,,,I

,

100

1001

f (Hz) 91

'

'

' ' ' ' 1

90

onset

/

89

H//ab

O o--O~

SQUID

I-88

\

peak

onset

H//c

[]

imp

--'

eak

87 0.01

w~

0.1

[Lt0hac (mT)

Fig. 14. Irreversibility temperatures as a function of (top) frequency and (bottom) amplitude of the ac ripple field. The open symbols denote the quasistatic conditions of SQUID magnetometry; for details cf. the text (Frischherz et al. 1995).

1991) by sample-dependent combinations of size and critical current density (according to the Bean model), i.e., the central issue is the increase of Jc with increasing magnetic field (lower panel). Such an increase of Jc with increasing field was occasionally seen in low-temperature superconductors, and at least three different mechanisms were identified, depending on the nature of the defects. The first mechanism prevails wiaenever an ordered defect structure occurs, i.e., the defect spacing d is constant throughout the superconductor. Then, whenever d equals the lattice parameter a0 of the flux-line lattice or multiples of a0, "matching" occurs and peaks in the pinning force or the critical current density are found. A nice example of these matching peaks was reported by Hillmann and Hauck (1972) for ordered precipitates in NbTi. In view of the underlying mechanism, the peaks always occur at the same induction B (which determines the lattice parameter) and are,

FLUX PINNING

207

2,£

,

,

i

1.5

~O

'

,

,

i-

.

i1~11

>"

0.5

I/

%~mm

-1.0

<

mmjfnmlnm~mm~mrma*mu

-I.5 ~

I 0

0

l

20 K HI[c

!

0.5

#00 hE

,

nn~nm-nn

n

2

,

'

,

I

4 poH (T)

'

i

,

i

,

,

6

n- ¸

i

8--

,

,

'

r

'

20 K HIIc

o

''J'''b'' (a)

'4''' Field (T)

~''

,

n

1

12

(b)

,

n.,

2

i

3

,

L

4

,

I ~

5

p

6

,

I

7

,

n

,

8

B (T)

Fig. 15. Fishtail anomalyin 123 single crystals: (a) magnetisationloop showing a characteristic dip (D/iumling et al. 1990): (b, top panel) magnetisationloop without a dip and (bottompanel) Jc(B) dependence evaluated from the data of the top panel (Werner 1997). therefore, temperature independent. Another mechanism was experimentally investigated by Evetts and Wade (1970), who studied a superconductor with a well-defined number of superconducting inclusions of (slightly) different superconductive properties. Since the elementary pinning force is, in this case, proportional to the difference between the reversible magnetisation of each compound, AMrev, the pinning force will decrease with magnetic field, go to zero (when AMrev = 0; i.e., where the two reversible magnetisation curves cross) and increase again at higher fields before finally dropping to zero near He2 of the matrix. This mechanism leads to a dip in the magnetisation curve followed by a broad maximum. Since the reversible superconductive properties of both phases depend on temperature, the peak itself will also be temperature dependent, in contrast to the first case. The third mechanism, which often leads to sharp peaks near He2 in nearly reversible superconductors, is related to the smallness of c66, the shear modulus of the flux-line lattice, near H~2, and can be understood in terms of collective pinning theory. A particularly complete study of this effect was made by Meier-Hirmer et al. (1985) using radiation-induced point defects in V3 Si single crystals, which varied in their concentration by more than three orders of magnitude. As shown in fig. 16, sharp peaks develop near Hc2 in the initial state and at low defect concentrations, which then start to broaden and to shift to lower fields prior to disappearing at high defect concentrations. Very good agreement with the static collective pinning theory (Larkin and Ovchinnikov 1979) was found at low defect concentrations, whereas the Jc-B dependence at'the highest defect concentration was attributed to plastic flow of the flux lines (Kramer 1973). According to these considerations, the first model for the "fishtail" anomaly in HTS single crystals (Dfiumling et al. 1990) was based on the assumption that oxygen-

208

H.W. W E B E R 10"1

u ,'3' 0 0 0 0 D O @ @ @Q @@@@

©' •

o

~)

• • 2,1022

' I~

i ~,

,

i

,

i

.

5 * 1 0 2 q n -2 0 0

0

• • iiiiii 10 9

2"10211"1021~ []

[]

[]~ D=~ ~ ~A~~

H[]

108

H[:lm o i-1

[]

m

[]

A

A • 5*10201,

i

107

A

•

A

A

ak

A

AAAAAAA

2 * 1020

I

/x

,,x

.:..o

10 6

0

0

0

O

0

0

1"1019 O

o

~' O

0

o

00© 0

0

l0 s 0

t

t

i

t

t

2

4

6

8

10

12

B (T) Fig. 16. Dependence of the persistent currents on magnetic field in a V3Si single crystal as a function of neutron fluence (Meier-Hirmer et al. 1985).

deficient parts of the superconductor with distinctly lower transition temperatures into the superconducting state (cf., e.g., the Tc-6 phase diagram of Y-123, Jorgensen et al. 1988, 1990, Cava et al. 1990, Breit et al. 1995) were the responsible defects, which would turn normal conducting at a certain temperature and magnetic field far below the transition temperature Tc of the single crystal and would, thus, be "switched on automatically" by the increasing magnetic field along the magnetisation loop. In addition, the appearance of "granularity" at higher magnetic fields had to be invoked in the early studies (D/iumling et al. 1990, Zhukov et al. 1993, Cohen et al. 1993), an argument that might have been applicable at that stage, but was later contradicted by transport measurements (Gordeev et al. 1994) showing a "fishtail" effect of Jo, and by magnetooptical studies of the flux penetration patterns (cf., e.g., fig. 8) in single crystals of better quality. Moreover, de Groot et al. (1991) pointed out that the fishtail anomaly did not disappear after extended oxygenation treatments, an observation that shed doubt on the entire idea of this mechanism being responsible for the fishtail. As a consequence, alternative models focusing mainly on the dynamics of the flux-line lattice, possible transitions between single-vortex pinning or pinning of small versus large bundles, and possible transitions between two- and three-dimensional vortex structures were proposed. (Actually, a very similar feature in the magnetisation curve of strongly two-dimensional HTS, such as Bi-2212 or T1-2223, which is sometimes called the "arrowhead" feature, might be caused by field-induced decoupling of pancakes, cf. sect. 2, and therefore not be related to the fishtail in 123 HTS, cf. also below). For instance, Krusin-Elbaum et al. (1992) and Civale et al. (1994a) studied magnetisation curves and relaxation, analysed their data in terms of a transition between a single-vortex pinning

FLUX PINNING

209

regime and collective pinning of vortex bundles, and included the collective creep concept (Feigel'man et al. 1989, 1991). The corresponding normalised relaxation rate, S, looked like the mirror image of the magnetisation curve (or the J s - H dependence) and the initial drop of the magnetisation was, therefore, attributed to the vortex dynamics, i.e., faster relaxation in the low-field regime, which differ under different pinning scenarios. These findings were soon questioned (Zhukov et al. 1994, Yeshurun, et al. 1994) based on careful measurements of the field dependence of S. These authors pointed out that the minimum of S was usually observed at fields significantly below the peak field of the fishtail, i.e. S was not the mirror image of M, and that a qualitatively similar relaxation behaviour occurred in crystals with and without the fishtail feature. The alternative explanations, however, differ, although both aim at a static origin of the anomaly in Y-123. Zhukov et al. (1994) favour changes in the elastic properties of the flux-line lattice and/or decoupling of flux lines into pancakes, which could also be associated with a decrease of the collective pinning correlation length resulting in an increase of Jc. (Decoupling into pancakes above 40 K would be inconsistent with data by Schalk et al. (1994), which were obtained on Y-123 thin films). Based on comparisons with Bi-2212, Yeshurun et al. (1994) pointed out that the "'effective" time scales accessible to experiment differ largely between the two materials (because of their strongly different collective pinning activation energies Uc, Ginzburg numbers and depairing current densities; cf. sect. 3), and that the anomaly would still be present in YBCO for t = 0, but not in BiSSCO. Evidence for the dynamic nature of the fishtail was obtained by van Dalen et al. (1995), who employed torque magnetometry on twin-free Dy-123 single crystals and evaluated their data in terms of the so-called general inversion scheme ("GIS", Schnack et al. 1993). The latter is based on the assumptions that the temperature and the current dependence of the activation energy can be separated and that U~ and the true critical current density Jc are connected by a power law. Such an analysis showed a smoothly decreasing dependence of the true J~ with field, which was interpreted as a clear sign for the dynamic origin of the fishtail; i.e., field-dependent relaxation rates and reduced activation energies at low fields. From the relatively weak field dependence of Jc, the authors conclude, furthermore, that the single-vortex pinning regime always prevails and no cross-overs occur. The most recent dynamic picture to be presented comes from measurements of the time evolution, OB/Ot, of the local induction with Hall probe arrays (Abulafia et al. 1996), which can be analysed in terms of the vortex velocity v (v = vo e x p [ - U / k T ] ) and, hence, in terms of a pinning potential U / k T without further assumptions about the creep mechanisms. The authors found a characteristic change in the slope of U / k T versus J at the peak field. This cross-over was related to a change in the creep mechanism at the peak field; i.e., collective creep prevails below the peak field, but a dislocation-mediated mechanism of plastic creep (Geshkenbein et al. 1989, Hirth and Lothe 1982) takes over at higher fields. With the advent of a new generation of Y-123 single crystals of extremely high purity, and in view of a thorough study of the fishtail and its angular dependence in several 123 single crystals, the original proposal by D/iumling et al. (1990) of oxygen-deficient clusters being responsible for the enhancement of Js received strong

H.W. WEBER

210 15

10

E <

m~

n,m~lm I ~

×

-5

Oxgen t r e a t m e n t : •

650°C 12.5h 115bar

×

-6

+ 510 °

-4

0.5 h

-2

1bar

0

poll (T)

2

4

6

Fig. 17. Reversible introduction and removal of the fishtail anomaly in Y-123 single crystals by oxygen treatments (Erb et al. 1997).

experimental support. In the first case, the application of the new (non-reactive) crucible material BaZrO3 (Erb et al. 1995, 1996a) and high-pressure oxygenation at 100bar (D/iumling et al. 1996) led to a complete disappearance of the fishtail (Erb et al. 1996b). Moreover, low-pressure post-ammaling followed again by the initial oxygenation procedure allowed to reversibly introduce and take out the anomaly, as shown in fig. 17 (Erb et al. 1997). Similar treatments of less pure single crystals failed. Consequently, the authors argue that in the high-purity case and under optimal doping conditions (6=0.09), two different situations prevail depending on the oxygenation process. Upon high-temperature/pressure oxygenation, the remaining oxygen-deficient unit cells are statistically distributed, whereas clustering of these unit cells occurs under lowtemperature/pressure annealing conditions due to a smaller entropy term in the free energy (Vargas and Larbalestier 1992). These clusters will then have certain superconductive properties (depending on their local value of 6), can be made normal conducting by certain magnetic fields - as described in the beginning of this section - and lead to the fishtail. On the contrary, clustering will always occur in less pure single crystals near and around the local impurity atoms, thus preventing the removal of the fishtail even under high-temperature/pressure oxygenation conditions. Of course, details of the above

FLUX PINNING

211

10~

~'-" <~ 10'~

10a

YBa2Cu307

100~--

?

'

z~" '

~--

B (T)

2.0 6 = 0.03

~i.5 < o

1.0 0.5 0.0

0

2

4

B(T)

Fig. 18. Dependence of the fishtail anomaly on (top panel) the global oxygen deficiency 6 and (bottom panel) oxygen clustering due to the finite mobility of oxygen at room temperature (Ktipfer et al. 1996).

arguments will have to be supported by further experimental evidence, but this is just a matter of time considering the recent pace of these developments. The second piece of evidence for oxygen-deficient lattice cells being responsible for the fishtail anomaly comes from a careful investigation of the angular dependence of the magnetic moments in twin-free and twinned single crystals at various levels of the oxygen-deficiency parameter 6 (Kfipfer et al. 1996, 1997). Leaving the influence of the twins aside for the moment (which lead to their own Js peak, cf. sect. 6.2) and concentrating on the results in twin-free crystals, both the development and the enhancement of the fishtail anomaly with increasing (5 were reported as well as a substantial increase of the peak height after extended annealing at room temperature (fig. 18). The first was explained by the increasing number of oxygen defects, the second by small-scale diffusion processes in the oxygen system, which favour the formation of clusters. Since the number of defects is a crucial parameter in the static collective pinning theory and because of the full experimental analogy to the above-mentioned study of point pinning centres in V3 Si single crystals (cf. fig. 16), the fishtail is proposed to be due to collective pinning by these oxygen-deficient clusters followed by plastic vortex interaction above the peak field. For further details regarding the discussion of relaxation and Jc(B,T) scaling we refer to the papers by Ktipfer et ~1. (1997) and by Meier-Hirmer et al. (1985). That the global oxygen deficiency is rather unimportant, but local clustering essential, is also confirmed by a systematic investigation of Y-123 single crystals with Tc's ranging

212

H.W. WEBER

from 14.0 to 92.8 K (Brandst~itter et al. 1996, 1998a). Fishtails are found in all cases (maybe with the exception of the crystals with a Tc of 14 and 21.8 K). If we consider crystals with very similar 6, e.g., with To's of 92.8, 92.1 and 90.2 K, the peak fields move from 3.5 to 9.5 and back to 4 T and the corresponding Jc's are 1.5, 1.8 and 2.6× 109 A m 2 (H ][ c, T = 50 K), respectively, i.e., clearly more dependent on the local arrangement of the oxygen atoms than on the global 6. In addition, the cross-over from 3D flux lines to 2D pancakes, which can be observed from the appearance of a crossing point in the re(T) curves or from the functional dependence of the irreversibility line on temperature, does not substantially alter the fishtail effect. To conclude this section, results on substituted 123 superconductors will be briefly summarised. Firstly, it turned out that due to special processing techniques under reduced oxygen pressure very high To's and sharp transitions were obtainable in light-R substituted superconductors, such as Nd, Sm and Eu and mixtures thereof (Yoo et al. 1994, Murakami et al. 1994, 1996). Both single crystals and melt-textured materials were produced in this way, but the details of the processing conditions depend on the substitution element in a critical way (e.g., Wolf et al. 1997). Partly because of its record Tc of ~ 9 6 K and partly due to the enormous fishtail effect, Nd-123 shows exceedingly high irreversibility lines, the irreversibility field (H [1c) at 7 7 K reaching ~13T (Wolf et al. 1997). As to the nature of the pinning mechanism responsible for the fishtail anomaly, mostly static pinning scenarios have been proposed. Whereas Wolf et al. (1997) analyse their data much in the same way as described above for Y-123, and ascribe the fishtail to oxygendeficient clusters (in addition to considering pinning by Nd-422 precipitates and by twin boundaries), Chikumoto and Murakami (1996) and Murakami et al. (1996) propose a N d - B a anti-site exchange mechanism leading to Nd-rich regions with diameters of 2 0 - 5 0 m n embedded in the Nd-123 matrix with a chemical composition close to 123. The clusters [Nd(Nd,Ba)2Cu307 ~], which were observed by TEM (Egi et al. 1995) and by STM (Ting et al. 1997), are then assumed to have weaker superconducting properties and to turn normal at the peak field. Only small influences of the size and concentration of 422-precipitates on the peak field are found (Yoo et al. 1995). However, the introduction of artificial pinning centres into melt-textured Nd-123 through neutron irradiation (cf. sect. 6.3), led to a complete disappearance of the fishtail peak (Weber 1996), i.e., to a smooth Jc(B) dependence, with low-field critical current densities near 2 x 109 A m -2 as compared to 1.5x 108 A m 2 prior to irradiation (H ]] c at 77 K). Similar trends had been observed before (Sauerzopf et al. 1995) on irradiated Y-123 single crystals both for H II c and for H II a,b. An additional feature, i.e., the recent observation of temperature-independent second peaks at low fields (around 2T) in Nd-123 single crystals processed under reduced oxygen pressure (Murakami et al. 1996), confirms the complicated interplay of different flux-pinning mechanisms to the measured field and temperature dependence of Jo (in agreement with the observations of Ktipfer et al. 1997 and Wolf et al. 1997). In summary, oxygen-deficient clusters seem to play a crucial role for the understanding of the "fishtail" effect in 123 HTS, although additional micro-structural work will be required for a more quantitative analysis of the flux pinning and depinning mechanisms.

FLUX PINNING

213

6. Flux pinning by artificial pinning centres The optimisation of the defect structure for flux pinning has, of course, been a major goal since the discovery of HTS. However, this turned out to be extremely tedious, in particular if metallurgical (processing) procedures were to be employed. Remarkable exceptions are the formation of normal-conducting inclusions in melt-textured R-123, such as the 211-precipitates in Y-123 or the 422-precipitates in Nd-123, or other additions such as CeO2, Pt, or MgO. Extended oriented planar defects are twin boundaries in all forms of R-123, dislocation chains (which seem to form naturally during the growth of R-123 thin films), and arrays of stacking faults, which form predominantly in melt-textured 123 materials. Most successful, however, was the application of radiation techniques, where strongly pinning defects could be achieved by certain interactions of radiation with matter, such as spherical statistically distributed defect cascades (fast neutrons), oriented arrays of cylindrical amorphous defects ("columnar tracks", heavy ions of very high energy) or randomly oriented cylindrical fission tracks (GeV-proton-induced fission in Bi- and Hg-based materials or thermal-neutron-induced fission of U additions to Y-123 and Bi-2223). The nature and, if available, the most likely pinning mechanisms of these defects will be reviewed in this section, while the impact of these defects on the critical current densities and the location of the irreversibility line will be mostly presented in sect. 7.

6.1. Normal-conducting inclusions Several reviews of the subject have been prepared (e.g. Murakami 1992, Salama et al. 1993, McGinn 1995, Sandiumenge et al. 1997), which do not only address the formation of the 211-particles ("green phase") in melt-textured 123 materials, but also the complexity of the defect microstructure, in general, which consists of twin planes, stacking faults, micro-cracks, sub-grain boundaries, minority phases etc., all coexisting in the matrix and each contributing to the bulk pinning properties in a different way. The 211-precipitates themselves (Jin et al. 1988, Zhu et al. 1991) were initially found to be very large (with diameters on the order of 20 ~tm), which required modifications of their growth conditions to render them suitable for flux pinning (2~ is only ~4 nm at 77 K, H II c, in Y-123). A description of various processes developed to achieve the required refinement of the size distribution of the precipitates can be found in the above review articles. An example of recent results obtained on directionally solidified Y-123/Y-211 composites is shown in fig. 19 (Sandiumenge et al. 1997). In this case (material with 15 wt% 211 precursor and 1 wt% CeO2 addition), the sizes follow a log-normal distribution with the position of the maximum at 0.4 ~tm; i.e., a factor of 50 down from the first results quoted above. Next, we have to address the question of whether or not these precipitates are directly responsible for flux pinning and, if so, through which mechanism. Normal-conducting precipitates are, of course, the first choice for strong pinning centres (cf. sect. 3), but their size should be comparable to the coherence length. From earlier work on low-temperature superconductors (Coote et al. 1972), we know, however, that sharp interfaces between

214

H.W. WEBER 40 35 30

O 25

20 ,.O 15

O 10 0

0.0

0.5

1.0 Mean

1.5

2.0

diameter

2.5

3.0

D (gin)

3,S

4.0

Fig. 19. Log-normal distribution of precipitate sizes in a directionally solidified Y-123/Y-211 composite; the position of the maximum is found at 0.4 p~m (Sandiumenge et al. 1997).

the matrix and large precipitates will lead to a substantial elementary pinning force, since the order parameter varies abruptly over a short distance and thus leads to a steep gradient of the pinning potential, which determines the pinning force. Consequently, surface pinning at the matrix-precipitate interface was proposed to be responsible for the high critical current densities achievable in these materials (Murakami et al. 1992). Employing in addition the direct summation of elementary pinning forces, which is reasonable in the case of strong pinning, the resulting critical current densities are proportional to Npd 2 (where Np is the number of precipitates per unit volume and d is their diameter). The latter is equivalent to V/d, where V ~ Np d 3 is the volume fraction of the 211-phase, and represents the effective surface area between the precipitates and the 123 matrix. Thus, by changing the volume fraction and the average size of the precipitates and plotting the resulting Jc versus V/d, the pinning mechanism can be identified (Murakami 1992), as shown in fig. 20 (Sandiumenge et al. 1997). Very good agreement with the predicted V/d dependence is found at relatively low fields (<1.5 T) and high temperatures, i.e., exactly under conditions where we expect a predominance of single-vortex pinning by strongly pinning defects. Indeed, other pinning centres and/or other flux-line configurations become more dominant in other field and temperature ranges, as discussed by Sandiumenge et al. (1997) in much detail. Returning to the pinning action of the interfaces, some concern was expressed about the validity of this argument, since phase boundaries tend to attract impurities, which could then in turn be responsible for flux pinning, e.g., through the Ag effect. However, the corresponding TEM images (e.g., fig. 21) did not provide any evidence for impurity agglomerations or strain at the interfaces. Lastly, it could be argued that the large 211-precipitates, which are readily observed by polarised light microscopy and are used for assessing the size distributions, such as in fig. 19, might be accompanied by smaller-sized, say 30nm, particles, which are not seen in the microscope, but would be suitable for pinning directly via the core interaction rather than via the surface pinning mechanism. This would, however, lead to a different dependence of Jc on V/d.

FLUX PINNING

1.2 109

O ..''"

45K .... • .... 6 0 K

o ...

.... ~ - - - 7 7 K "~

215 O

..--

8 10 8

E

.-"

•

4 IOs . _--"

•

0 0

D

.............. ~'"

,,IL~*,I,,,,l,,,ll,l,llljp,l~,,J

500

1000

1500

2000

2500

3000

3500

V / d ( c m -1) Fig. 20. Jc versus the effective surface area of Y-211 precipitates m the t23 matrix. The observed linear dependence at low fields and high temperatures confirms single-vortex pinfling at the interfaces and direct summation (Sandiumenge et al. 1997).

Fig. 2l. High-resolution TEM images of211-123 interfaces (from Murakami 1992, p. 92). Note that the interfaces are clean.

216

H.W. WEBER

6.2. Extended planar defects In this subsection, I will discuss defects such as twin boundaries, stacking faults, dislocations, dislocation chains, and nano-rods, which are not always introduced artificially, but sometimes develop naturally during the growth of the 123-phase (such as twin boundaries in all forms of 123 materials, stacking faults in melt-textured materials, and dislocation chains in epitaxial thin films). Again, a huge number of papers has been devoted to each of them, from which I will select a few for demonstrating their flux-pinning effect. The first and most widely investigated defect certainly is the formation of the twin boundary structure in the R-123 system, which is caused by the necessary phase transformation from the tetragonal to the orthorhombic crystal structure during processing and serves the purpose of relieving the resulting strains by occasional interchanges of the crystallographic [100] and [010] directions. The resulting {100} twin planes are extended planar defects, which go right through the whole thickness of single crystals along the c-direction, but may be arranged in completely different ways depending on growth conditions. In addition to their obvious dependence on the oxygen content (the orthorhombicity becomes smaller with increasing 6), influences of certain substitutions, e.g., Fe doping (W6rdenweber et al. 1989), or of the presence of 21 l-precipitates (Miiller and Freyhardt 1996), were reported. Two examples obtained by polarised-light microscopy on Y-123 single crystals are shown in fig. 22, one referring to a dense cross-linked pattern, the other to an arrangement of a few parallel twins. Extensive structural investigations (e.g., Zhu 1995) clearly demonstrate structural disorder associated with each boundary, which will lead to flux pinning along the entire length of the boundary, i.e., an extended planar defect, which will prevent flux motion across it. Evidence for flux pinning by twin boundaries came predominantly from transport measurements on single crystals and thin films. A nice example, reported for a single crystal with preferential twin boundary orientations along the [110] and the [110] directions (Kwok et al. 1990a, 1991a), is shown in fig. 23. Here, the magnetic field is kept parallel to the a,b plane and the sample is rotated in such a way that the transport current I is at certain angles 0 with respect to the field direction. The angle 0 = 0°, therefore, corresponds to a force-flee configuration with H [[I, where the Lorentz force

Fig. 22. Twin patterns in Y-123 single crystals, as observed by polarised light microscopy(Werner 1997): (a) dense cross pattern; (b) individualunidirectionaltwins.

FLUX PINNING

217

44 T=91.88

42

K

~

_

_

H=1.5 T

40

X

38

%

36

~

H±I

A

-ill 2

34

12

10

C3_

6 4

-

T=91.26

[

K

: tt IItwin planes

2 1 0 -70

0

70

o (o)

140

210

Fig. 23. Angular dependence of the resistivity in an Y-123 single crystal at three temperatures. The circles in the middle panel show the onset of pinning by the twin planes, which becomes fully developed in the bottom panel (Kwok et al. 1990a).

goes to zero leading to a minimum in the resistivity p (0 = 0 °, 0 - 180°). Sharp decreases of p, about 2 ° wide, occur at angles o f - 4 5 ° and 45 ° from the maximum (0 = 90°), i.e., whenever the field is aligned parallel to the twin planes ([110] and [1 TO]) and flux pinning by the twin boundaries becomes effective. However, details of this angular dependence were soon found to be sample dependent (Kwok et al. 1990b, 1991a), i.e., influenced by the presence or absence of additional defect structures. Transport measurements of Jc on Y-123 thin films (Roas et al. 1990) provided evidence for flux pinning by extended planar defects, which were assumed to be twin planes or stacking faults. In these experiments, the condition HA_I, i.e., the maximum Lorentz force, was always maintained and the sample rotated under this condition from H [[ a,b ( 0 = 0 °) to H It c (0=90°). The corresponding angular dependence of Jo shows extremely sharp maxima for H [[ a,b, which are attributed to the intrinsic pinning effect (Tachiki and Takahashi 1989), but a broader second m a x i m u m for H [[ c, which can only be explained by the presence of extended defects growing along the c-axis. Contrary, to the above assumption of twin planes being responsible for pinning in the case H [] c, Schalk et al. (1996) proposed a novel extended defect structure, i.e., dislocation chains along the c-axis, based on a thorough study of the growth conditions following various processing routes in Y-123 thin

218

H.W. WEBER

8 _5=0 . 0 / 9 ~ E 2

~'-'6

E

fi4 o

NdBa2Cu307

8-

2

0

2

B(T)

4

6 (a)

0

2

4

6 B(T)

8

10

12 (b)

Fig. 24. Twin boundary pinning in Y- and Nd-123 single crystals as deduced from magnetic measurements: (a) Angular dependence of Am at various angles near to the c-direction; note that twin boundary pinning (PEtw) disappears for misalignment angles above 8° (Ktipfer et al. 1996). (b) Persistent currents as a function of field for various oxygen deficiencies 6; note that PEtw disappears for 6 > 0.08 (Wolf et al. 1997).

films. The experiments consisted of employing scanning tunnelling microscopy (STM) to determine the growth patterns first and then of measuring transport Jc's on patterned parts of these samples to assess their superconductive properties and to establish a correlation with their microstructure. In contrast to the generally small density of growth spirals and associated screw dislocations (Gerber et al. 1991, Mannhart et al. 1992), which are usually found to be on the order of 1012 m -2, the dislocation chains, formed as a consequence of the cluster-like deposition and the two-dimensional growth mode, can be spaced as closely as a few 10 nm apart, thus leading to higher defect densities by about three orders of magnitude, which renders them suitable for flux pinning, even at high magnetic fields, and additionally results in similarly broad peaks for H II c, as found by Roas et al. (1990). The above number of 1015 dislocations per m 2 has recently been confirmed experimentally in a TEM study of highly bi-axially oriented Y-123 thin films (Svetchnikov et al. 1997). Returning to flux pinning by twin boundaries, the interplay of this mechanism with that exerted by other defects will be discussed on two examples. First, transport measurements on twinned and detwinned single crystals clearly showed the interplay with intrinsic pinning and the "lock-in" transition (Feinberg and Villard 1990) of the flux lines in the magnetoresistivity close to Tc (Kwok et al. 1991b). From the differences in the angular dependence of the resistivity drop (the one related to intrinsic pinning is extremely sharp, ~0.3 ° wide), the authors were able to determine the onset temperatures and fields in the vicinity of Hc2, at which each of these mechanisms becomes effective. Magnetic measurements on single crystals and melt-textured Y- and Nd-123 (Ktipfer et al. 1996, 1997, Wolf et al. 1997) address the interplay of twin boundary pinning, the fishtail anomaly (cf. sect. 5) and/or intrinsic pinning. From measurements of the angular dependence of the hysteresis widths Am of the magnetisation curves, the authors were able to define a second peak (or a shoulder) at lower fields (fig. 24), which is most pronounced for H ]] c, i.e., the currents flowing in the a,b plane, and disappears for misalignment angles above ~8 °. Such peaks are observable only when the

FLUX PINNING

219

Fig. 25. TEM image of stacking faults near 211-precipitates in melt-processed Y-123 (Zhang et al. 1995a). competing pinning interactions become small, e.g., collective pinning by point defects of relatively low concentration at higher temperatures. Furthermore, these twin boundary peaks disappear with increasing oxygen deficiency 6, which was explained by the increasing strength of collective pinning due to the enhanced number of point defects, i.e., statistically distributed oxygen-deficient lattice cells. In summary, the question of whether or not twin planes are strongly pinning defects, is certainly not easily answered. In my opinion they are not, because they were shown to be effective only for flux lines closely aligned to preferentially oriented twins (otherwise, only the crossing points would act as pinning centres). Theoretical estimates of the role of twin-boundary pinning in melt-textured materials support this from a practical point of view (Fujimoto et al. 1992). Indeed, twin planes could actually serve the opposite purpose by providing "channels" of easy vortex motion (Oussena et al. 1995) and thus eventually develop into a "weak link" at higher magnetic fields, as was recently observed by the magneto-optical technique on a single crystal containing only one major twin boundary (Werner 1997). The next type of extended planar defects to be discussed are stacking faults and dislocations, which play a major role in melt-textured materials. TEM observations of stacking faults in Y-123 (Zandbergen et al. 1988) proved them to basically consist of CuO double chain layers. Their formation can proceed under different mechanisms (Rabier et al. 1993, Gu et al. 1993, Zhang et al. 1995a, Sandiumenge et al. 1995), which depend again on details of the processing conditions (such as temperature and oxygenation procedure) and can lead to large arrays extending over several tens of nm to several ~tm. An example of such an array is shown in fig. 25 (Zhang et al. 1995a). In addition, dislocations are generally found to be associated with the stacking faults. Dislocations are otherwise introduced into melt-textured materials artificially, e.g., by hot isostatic pressing or by high-temperature uniaxial compression (SelvamanSckam et al. 1993, Martinez et al. 1996), and into single crystals by high-temperature indentation (Yoshida et al. 1990). In all cases, the high temperatures needed for these special processing conditions severely affect the "as-grown" properties of the superconductors, i.e., require post-processing

220

H.W. WEBER i

u

i

;

i

L

i

--D--PMP1 77 K " - - R - - R e - a n n e a l e d PMP1

10 9

$

l0 s H II ab

~

_

I~oH (T)

Fig. 26. Field dependence of the critical current densities in melt-processed Y-123 in the as-grown and in the re-annealed state. The reduction of Jo is related to the reduction of the number of stacking faults by three orders of magnitude (Zhang et al. 1995b).

oxygenation treatments, which in turn affect the newly introduced defect structures. Hence, an analysis of the flux-pinning action of these defects is extremely complicated. Flux pinning by dislocations has been well established in the metallic superconductors (Kramer and Bauer 1967, Freyhardt 1971b, Dew-Hughes and Witcomb 1972, cf. also the monographs by Campbell and Evetts 1972 and by Ullmaier 1975). The elementary pinning force critically depends on the type of dislocation (screw dislocations parallel or perpendicular to the flux lines, edge dislocations, etc.). Core pinning by dislocations in melt-textured HTS was recently considered by Ullrich et al. (1993), who calculated the pinning potential under the assumption of a line defect (dislocation plus the surrounding strain field) with a reduced order parameter and employed direct summation with the dislocation density Nd. As pointed out above, no "model" systems are available for studies of flux pinning by these defects on their own, since they are always accompanied by other strongly pinning defects. However, Zhang et al. (1995b) reported on a special annealing experiment of "powder-melt-processed" Y-123 (Zhou et al. 1990), which allowed stacking faults and most of the dislocations to be removed without nominally changing the remaining defect structure. The corresponding reduction of Jc, which was also accompanied by reductions of the irreversibility lines for both field orientations, is shown in fig. 26; it was explained by the strongly reduced number of stacking faults (from 106 cm -1 to about 103 c m 1 according to TEM images). Finally, I wish to mention the possibility of introducing artificial extended objects, such as nano-tubes or nano-rods, although this has not been applied to 123 materials so far. The first report came from Fossheim et al. (1995), who embedded carbon nano-tubes into ceramic Bi-2212 superconductors and found some enhancements of flux pinning from magnetisation measurements. More recently, MgO nano-rods were introduced into Bi-2212, T1-2223 and Bi-2223 (Yang and Lieber 1996, 1997), which seem to be advantageous because of their chemical inertness and which allowed good texturing of the superconductors.

FLUX P1NNING

221

6.3. Radiation-induced defects Extended defects can be produced in all classes of HTS by radiation techniques, if the interaction of the particles with the constituents of the superconductor is sufficiently strong, i.e., the primary knock-'on atom has energies above ~1 keV in the case of elastic nuclear interactions, or the interaction occurs electronically and leads to amorphisation of the material over extended areas. I will discuss in the following the most important types of radiation suitable for the production of extended defects and the nature of these defects, which was established in most cases by TEM. The first indication of strong flux-pinning enhancements came from fast neutron irradiation of Y-123 single crystals (Umezawa et al. 1987). Since the samples are exposed to this kind of radiation in a reactor, the experimental conditions for an assessment of the neutron-induced defects were quite difficult. Therefore, transmission electron microscopy was employed in two stages: first by simulating the defect production by fast neutrons through in-situ irradiation with suitable ions (e.g., Kr or Xe with an energy of 50 or 85keV), and second, by directly exposing TEM samples to various reactor neutron spectra and fluences (Frischherz et al. 1993, 1994). The results may be summarised as follows. Collision cascades were identified to be the major defect, with sizes ranging from below 1 nm (the resolution limit of conventional TEM) up to 5 nm, the average defect diameter being 2.5 nm. They are surrounded by an inwardly directed (vacancy) strain field, which renders the total defect size an approximately spherical volume with a diameter of 6 nm on average. The density of cascades was found to be 5 × 1022m 3 at a fluence of 1022 m -2 (E > 0.1 MeV) and to scale linearly with fast neutron fluence. They remained stable at temperatures up to at least 400°C. High-resolution images of simulated (Xe irradiation) or neutron-induced defects showed mostly amorphous, but in some cases re-crystallised, defect structures, thus confirming their suitability for flux pinning. A highresolution TEM image of such a defect is shown in fig. 27 (Weber 1996). As mentioned above, fast neutron irradiation is usually performed in fission reactors, which provide a characteristic distribution of neutron energies (cf., e.g., Weber et al. 1986). Fast neutrons with energies above 0.1 Mev, which are mainly responsible for cascade production because of their ability to transfer sufficient energy to the primary knock-on atoms, represent only a minor fraction of the neutrons interacting with the superconductor. These lower-energy neutrons create point defects (Frenkel pairs) and defect clusters of smaller size. Oxygen, being the lightest constituent of the system, is preferentially displaced, resulting in a certain additional disordering of the oxygen subsystem and, hence, in a depression of Tc. This effect and the influence of annealing steps, which lead to an enhanced mobility of the oxygen atoms, were reported by Sauerzopf et al. (1997); their results are shown in fig. 28. Accompanying model calculations (Trushin et al. 1996) revealed that vacancies in the CuO chains were mainly responsible for the observed effects. Hence, neutron irradiation leads in ge~neral to complicated interactions of the newly introduced defects with the pre-irradiation defect structure. This interplay between "new" and "as-grown" defect structures also plays a major role when critical current densities and irreversibility lines are analysed and related to flux-

222

H.W. WEBER

Fig. 27. High-resolutionTEM image of a fast-neutron-inducedcollisioncascade in Y-123 (Weber 1996). pinning mechanisms in 123-superconductors. From the size of the collision cascades, which matches almost exactly the flux-line diameter at 77 K, we expect strong pinning at elevated temperatures (through the core pinning interaction). This is confirmed (e.g., Sauerzopf et al. 1995) by the huge enhancements of Jo in 123 single crystals and the observed upward shifts of the irreversibility line (H II c). However, (low-energy) proton irradiation, which produces mainly point defects or small point-defect clusters, also results in considerable Jc enhancements (Civale et al. 1990), although without the accompanying change of the irreversibility line. Hence, the question of which defects contribute to what extent to flux pinning in the case of neutron irradiation needs to be addressed. This was done in several steps (Wiesinger et al. 1992b, Vlcek et al. 1992, 1993, Sauerzopf 1998), the basic idea being related to the differences in mobility between point defects and cascades upon exposing the material to annealing cycles at elevated temperatures (note that the cascades remain tmaffected up to temperatures of at least 400°C, cf. above, Frischherz et al. 1994). It turns out (Sauerzopf 1998), that the contribution to Jc(H IIe) of smaller defects is significant as long as the concentration of the strong pinning centres is small, i.e., at low neutron fluences, and can be completely removed by an annealing step at 250°C. On the other hand, the increase of Jc (after annealing, i.e., after removal of the additional point defects) is linearly correlated with the concentration of the cascades and becomes the only relevant contribution at

FLUX PINN1NG

223

n

92.6

0

?

.._.. 92.4 V v

¢,1

o

I-- 92.2

~1 ==/

92.0

"- •

i

i

i

6 o

495

r"

c'~

5.00

o

5.05

\

0 -~ed.

0 neutron

lx1021 fluence

o

2 x 1 0 ~1

(m -2, E > O . 1 M e V )

Fig. 28. Influence of neutron irradiation and annealing on the transition temperature T c of Y-123 single crystals. The lower panel shows results of model calculations, which relate the above changes predominantly to the formation and removal of additional vacancies in the CuO chains (Sauerzopf et al. 1997, Trushin et al. 1996).

cascade concentrations of around 2 × 1022 m 3. For H [Ia,b, i.e., when current flow along the crystallographic c-direction is involved, the situation becomes more complicated due to the presence of intrinsic pinning effects and due to changes of the material anisotropy caused by the introduction of scattering centres, i.e., the additional point defects. Another (and equally or even more) effective irradiation technique was introduced by several groups (Civale et al. 1991, Konczykowski et al. 1991, Gerh/iuser et al. 1992, Hardy et al. 1992) and is based on a completely different interaction mechanism with matter, i.e., the electronic energy loss of heavy ions of very high energy, e.g., 5 GeV Pb or 1 GeV Au ions (for a review, see Civale 1997). In this case, the damage process leads to complete disorder (amorphisation) of the material along the direction of the incident particles, i.e., to the formation of "columnar tracks". Hence, the defect production is completely different from that of fast neutrons, in particular also with regard to their arrangement; i.e., we find perfect alignment of the tracks along the particle direction as opposed to a statistical distribution in the case of fast neutrons. This situation is shown schematically in fig. 29 (adapted from Kraus et al. 1994a), where the interaction of charged particles with matter and the corresponding defect production are depicted for ions of different mass and energy. It should be noted that the production of continuous columnar tracks requires an electronic stopping power Se of at least 20 keV/nm (in Y-123): for lower Se oriented (nearly continuous) "strings of beads" are formed, and for still lower Se individual amorphous defect areas along the particle trajectory. As a consequence, the pinning action of the tracks is, in general, strongly anisotropic, and it is optimal for tracks oriented

224

H.W. WEBER

Particle Trajectory

t~

: i

I Se

... 5

...

6 MeV tie

12 .

.

.

.

.

.

2,5 ...

230 MeV Ni 25 MeV S

(keV/nm)

340 MeV Xe 173 MeV Xe

1 GeV Pb

Fig. 29. Schematicrepresentationof the defect formationupon ion irradiation. Se denotes the electronic energy loss (stoppingpower), the numbers are only approximate(adapted from Kraus et al. 1994a). parallel to the magnetic field direction. In this case, vast improvements of Jc and shifts of the irreversibility line were reported, which go far beyond those achievable with collision cascades. Further fundamental work that allows one to probe details of the magnetic microstructures, e.g., the question of forming flux lines or pancakes, was done on Y-123 and Bi-2212 thin films (Holzapfel et al. 1993, Kraus et al. 1994a) and should be mentioned here as an example of research on flux-line dynamics and the nature of the vortex-solid structures evolving under "correlated" disorder as provided by the columnar tracks (Civale 1997). Again, the defects were investigated carefully by TEM (e.g., Hardy et al. 1991, Zhu et al. 1993). It turns out that each particle leads to the creation of a track, which is roughly cylindrical in shape with a diameter on the order of 5 - 1 0 n m (fig. 30) and has an amorphous interior. I f the energy of the incident particles is below optimum, i.e., the stopping power falls below a certain material-specific value, the shape of the defect along the trajectory becomes inhomogeneous as shown schematically in fig. 29. In all cases, and even with the highest energies employed so far, the thickness accessible for these types of radiation-induced defects is limited to a few tens of gm. This is certainly the most serious disadvantage of the technique. The other drawback, viz. the preferential orientation of the tracks, can be at least partially removed by forming "splayed" tracks (Civale et al. 1994b), but the distribution will of course never be random and, therefore, never effective for arbitrary orientations of the magnetic field. The final possibility for introducing extended defects into high-temperature superconductors by radiation was proposed and tested on ceramics quite early on (Fleischer et al.

FLUX PINNING

225

Fig. 30. TEM images of heavy-ion-inducedcolumnar tracks in Y-123 single crystals. The top panel shows the amorphous (cylindrical) defect zone along the particle trajectory, the bottom panel views the defect from the top. 5.3 GeV Pb ion irradiation to a fluence of 6×1015m-2 (Hardy et al. 1991). 1989) and consists o f introducing fission tracks into the material, either by adding small amounts of fissionable nuclei, e.g., 235U, and exposing the superconductor to a beam o f thermal neutrons (Weinstein et al. 1993, 1994), or by direct fission o f suitable nuclei of the superconducting material, e.g., 2°9Bi or 2°°Hg, by very-high-energy protons (KrusinElbaum et al., 1994). In the latter case, 800MeV protons were employed by Krusin-Elbaum et al. (1994) to induce fission o f the 2°9Bi nuclei in Bi-2212 and -2223 tapes, and more recently of Hg nuclei in Hg-based 1212 ceramics. The corresponding fission products, such as Xe or Kr, have sufficient energy to produce columnar defects in much the same way as discussed above. However, due to the statistical nature o f the fission locations, the directions of the columnar tracks are random, i.e., the desired splay o f these extended defects is automatically achieved. TEM on such an irradiated Bi-2212 tape showed columnar defects with diameters o f 7 nm and lengths in excess o f 6 ~tm, which is found to be consistent with the expected total range o f columnar defect production by the fission tracks o f 7 ~tm for 8 0 M e V Xe and of 12 ~tm for 100Mev Kr. The results confirm the beneficial effects of these defects on the critical current densities, the irreversibility lines as well as the creep behaviour. Obviously, more work is still needed to assess the potential of this kind of radiation treatment. However, in view o f the enormous penetration range o f such highenergy protons into the superconductors, the most serious limitation of the track formation as discussed in the previous paragraph is clearly removed. The second possibility (Weinstein et al. 1993, 1994) is based on adding small amounts of fissionable nuclei, such as 235U, to the superconductor prior to processing, and to expose the final product to a beam o f thermal neutrons, thus inducing fission o f the added uranium and creating fission tracks in much the same way as discussed in the last paragraph. Thermal neutrons have a very large penetration range in all superconductors

226

H.W. WEBER

Fig. 31. TEM image of fissiontracks created by thermalneutron-induced fission of 235U admixtures to meltprocessed Y-123 (Weinstein et al. 1998). of interest and are, in addition, much more readily available and of beam sizes even suitable for processing of components, if desired. The fission products are well known in this case, they have a total energy of 160-180MeV and create columnar defects of isotropic distribution, again with a diameter of~10 nm and lengths on the order of 10 ~m. Due to the relatively low energy of the fission products, the defects usually have the character of the above-mentioned "strings of beads" (fig. 31, from Weinstein et al. 1998). In the specific case of melt-textured Y-123, detailed studies of the chemistry and the phase formation during processing revealed the formation of a stable U - P t - Y B a - O compound, which is dispersed in the matrix in the form of extended particles with sizes of ~300nm. Consequently, all of the fission tracks originate from these clusters and form three-dimensional "star-like" patterns in the matrix. However, if we consider their flux-pinning action, projections of these three-dimensional defect structures onto the plane perpendicular to the applied field, i.e., into the symmetry plane of the flux-line lattice, must be made, which turn out to result in effective defect sizes with diameters of 10-20 nm, i.e., quite similar to (about twice as large as) those created by fast neutron irradiation (Eisterer et al. 1998). Indeed, first results of the exposure of U-doped melt-textured Y-123 monoliths to thermal neutron irradiation confirm the similarity of the Jc enhancements to those obtained upon introducing collision cascades by fast neutrons. Some of these results will be presented in sect. 7. However, if we aim at the optimisation of large pellets, e.g., for maximum trapped-field conditions (cf. sect. 8), the sizes of these pellets may easily surpass the space limitations of fast neutron irradiation facilities, which renders U-doping followed by thermal neutron irradiation an unchallenged tool for this purpose (due to the much less stringent spatial restrictions in the latter case). Finally, I wish to point out that alternative suggestions for doping the superconductors, e.g., by adding Li to the material and relying on the Li(n,(0 reaction (Zhao et al. 1992), do not fall into the category of radiation treatments to be discussed in this context, since the resulting defects are too small to act as effective strong pinning centres. In summary, various possibilities of introducing artificial defect structures into all families of high-temperature superconducting materials have been explored during the past few years, only a few of them being effective for enhancing their flux-pinning

FLUX PINNING

227

properties at high temperatures and magnetic fields. Among them, some techniques are more suitable for fimdamental studies of the flux-line-defect interactions (fast neutrons, columnar tracks), others seem to be promising also from an applied point of view (fast neutrons because of the statistical defect distribution and the large penetration range, fission tracks for the same reasons, but in Y-123 only, if the preparational aspects, i.e., successful admixing of the doping elements without property degradation, can be solved, as has been demonstrated for Y-123). Both fast-neutron and thermal-neutron irradiation do not seem to represent a major problem for materials and applications which do not require a silver sheath, as far as the induced radioactivity of the superconductors is concerned. Silver, however, activates badly under a thermal neutron environment, which might make other radiation treatments, such as proton-induced fission, more attractive in this case.

7. Critical current densities and irreversibility lines

Published results on Jc and on the irreversibility lines in R-123 superconductors would presumably fill a few books as of today. Moreover, the caveats addressed in sect. 4 and our own experience with data on hundreds of samples prevent me from trying to chase for "record" Jc's and the "highest" irreversibility lines in the literature. I will rather choose a few examples, preferably on systems which allow the experiment to be done on the same sample prior to and following a certain treatment, in order to make sure that the defect structure for flux pinning remains otherwise (almost) unchanged. The point here is rather to show along which lines flux pinning can be optimised. Some of these aspects were already presented in the preceding sections, the main emphasis here will be to expand on the actual Jo's and the irreversibility lines. As the first example, I take the influence of oxygen deficiency. It is tempting to expect that (small) deviations from optimal doping will lead to additional point defects, which should add to flux pinning. In addition, the enormous change of the superconducting parameters upon reducing the global oxygen content (the anisotropy parameter increases by a factor of ~20 by reducing Tc from 92 to 20K, Chien et al. 1994) will make this system ideally suited for studying dimensionality effects on flux pinning. These issues were addressed in grain-aligned ceramics, thin films, single crystals and melt-textured bulk samples (Seidler et al. 1991, Feenstra et al. 1992, Gray et al. 1992, Ossandon et al. 1992a,b, Thompson et al. 1992, 1994, Jones et al. 1993, Wisniewski et al. 1994, 1996, Brandst/itter et al. 1996, 1998a, van Dalen et al. 1996, Samadi Hosseinali et al. 1997). The first conjecture is ruled out by all data on all forms of 123 superconductors, i.e., it turns out that Jc decreases immediately if c5 starts to deviate from its optimum value. This is illustrated in fig. 32, where Jo is plotted as a function of magnetic field in melttextured Y-123 superconductors (H I1c) with transition temperatures between 92.4 and 45.8K (Wisniewski et al. 1996). The data deliberately refer to very low temperatures (5 K in this case), in order to minimise the influence of variations in Tc. A very similar behaviour is found for the irreversibility lines (IL's) as shown in fig. 33 for H I[ c, i.e.,

228

H.W. WEBER

Hlrc 1 0 lc

10 9

O'-O"O-..O.,.O..

~ _. O

To(K) +"+"+"+"+--2:; -- O-- - [] ~ ~ ~ x ~ + ~

92.4 89.4 73,6 61.6 45.8

108

00H (T)

Fig. 32. Critical current densities versus applied field (at 5 K) in melt-textured Y-123 bulk superconductors (H r] c) with various degrees of oxygen deficiency (Wisniewski et al. 1996).

they start to move to lower fields and temperatures immediately upon changing 6. As suggested by Almasan et al. (1992), a scaling invariance of the 1Us can be found, if the irreversibility fields Hirr are normalised by a certain scaling field H +, which is defined as Hirr at T/Tc = 0.6 (fig. 33b). This shows that the data follow the power law Hirr =/3[ 1 - (T/Tc)]1.5 at high temperatures, indicative of flux-line depinning (Malozemoff et al. 1988), but at lower temperatures turn into an increasingly exponential behaviour with decreasing T~, which is characteristic of 2D magnetic microstructures. In fact, careful measurements of the magnetic moments on oxygen-deficient single crystals in the reversible regime show a crossing point of the m(T) curves at various fields, beginning with 6 = 0.32 (Brandst/itter et al. 1998a), which is considered to be the signature of vortex fluctuations in 2D superconductors (Bulaevskii et al. 1992, Koshelev 1994, Brandst/itter et al. 1997). However, even in the 2D regime, say around the 60 K plateau, a variation of 6 does not lead to enhancements of J~, i.e., the role of point defects is marginal also under these circumstances. (Note, however, that extended room-temperature annealing may lead to the formation of clusters due to oxygen mobility and contribute to flux pinning in this way, as discussed in sect. 5.) In summary, the change of flux pinning in O-deficient Y-123 seems to be primarily due to the gradual transition from 3D flux lines to 2D pancakes. In addition, as discussed in much detail by Thompson et al. (1994), the reversible mixed-state parameters (such as the condensation energy) degrade in a similar way as T~ resulting in a reduced pinning capability through this fundamental effect. An interesting analysis of critical current densities and relaxation rates in O-deficient thin films (van Dalen et al. 1996) allows the flux-pinning mechanism to be identified by employing the general inversion scheme (Schnack et al. 1993, cf. also sect. 5). The temperature dependence (fig. 34) of the (unrelaxed) critical current densities in several O-deficient systems (with 6 varying fi'om 0 to 0.4) can be described by defining a single disorder parameter in the framework of Ag pinning (cf. sect. 3) of 3D single vortices, where g is the mean free path of the

FLUX PINNING

229

15

10

5 ~ O0

20

40

(a) 80

1130

T(K) 10

i

,~

92.4

o X +

73.6 61.6 57.7

[] 89.4

~-~/ , o~ / _~Y'/

•zx 45.8 27.~/~

s¢

T/To = 0.6

0.1

(b) Hl~c

,

I

0.1

,

1 - TFFc

,

Fig. 33. Irreversibility lines in melt-textured Y-123 bulk superconductors (H II c) with various degrees of oxygen deficiency (Wisniewski et al. 1996): (a) Hin, versus temperature; (b) normalised Hit r versus [ 1 - (T/To)].

charge carriers. Alternative mechanisms, such as ATe pinning or pinning by individual oxygen defects, are incompatible with the experimental data. The second example is still related to the issue of point defects and will briefly describe two experiments. In the first, advantage is taken of the (scarce) availability of isotopically pure rare earths, which are then synthesised to form 123-ceramics. For instance, in the case of Gd, two isotopes can be used, 155Gd and 16°Gd, which have tremendously different absorption cross sections for thermal neutrons (a = 61.000 b and a = 0.77 b, respectively). Hence, if 155Gd is exposed to a thermal neutron beana, each thermal neutron will be absorbed and will produce a Gd recoil atom having a recoil energy of 180 eV. According to TRIM calculations (Biersack and Haggmark 1980) this recoil will result on average

230

H.W. WEBER

1.01 0.8

A

......... " ......... i

~

0.6 0.4

' •

"

Xn = 7,0

"'"'-, ~Tc.pinning V~

* xn : 6.85

"",

~

6J-pinning',,..

0.2 0.0 0.0

1,0,

0.2

•

0.4 0.6 t = T/r e

0.8

........... ........... "--'---.

0,8

1.0

B

" ............. " ""-, 6Tc-pinning

0.6 0.4 0.2 0.0 0.0

• x~=7,0 A Xn=6.7

0.2

•

Sl-pinning"',, • ~

0.4 0.6 t = T/Tc

'a,""",,

0.8

1.0

Fig. 34. "True" critical current densities in various oxygen-deficientthin films versus reduced temperature. The solid lines correspond to flux pinning due to spatial fluctuations of the mean free path and the dashed line to pinning due to spatial variations of the transition temperature (van Dalen et al. 1996).

in the formation of 2.7 point defects per Gd recoil, i.e. will lead to the formation o f randomly distributed Frenkel pairs (strictly point defects). On the contrary, a reference sample containing 16°Gd should be practically unaffected by the thermal neutron beam. Two sets o f experiments along these lines have been reported so far (Brezgunov et al. 1990, Sickafus et al. 1992), but both were not "clean" in a strict sense, because they did not employ "purely" thermal neutrons, but rather the full neutron spectrum with and without a Cd shield, i.e., all neutrons except for the thermals in the first case, and all neutrons in the latter. Hence, the contribution o f the high-energy neutrons (cf. sect. 6.3) had to be subtracted in some complicated way. A new experiment with an extremely well thermalised neutron beam (i.e. a Cd ratio o f about 500) was made over a period o f four years (due to the very low flux density resulting from a pure thermal neutron beam; Brandst~itter et al. 1998b) and confirmed the marginal role of point defects in an unambiguous way. Approximately 5x 1025 point defects per m 3 (an unknown amount o f them may have recombined immediately) were necessary to enhance Jc in 155Gd-123 by 50% (10K, 0.3 Y), and Jc has reached 2.2× 109 A m -2 (10K, 1 T) at twice this defect concentration, up by a factor o f ~2 compared to the unirradiated state. The ceramic sample containing 16°Gd did not show any changes o f Tc or Jc.

FLUX PINNING

E

~oH =1 T

.x .....

231

........... x"-I .-'

% 2

x--

HW01

i

/

i

i

i

0

1

2

3

4

neutron fluence (102~ m 2, E>0A MeV)

Fig. 35. Critical current density (H II c) at 77K of an Y-123 single crystal as a function of fast neutron fluencc. Thc crosses refer to sequential irradiation without annealing, the solid symbols to sequential irradiation prior to annealing, and the open symbols to the results after annealing (Sauerzopf 1998).

Similarly small effects on Jc were found upon irradiating Y-123 single crystals (at low temperature or room temperature) with 1 MeV electrons (Giapintzakis et al. 1992, 1994a,b) to very high fluences (up to 1024 m-2). In this case, mainly O and/or Cu atoms from the CuO2 planes are displaced, leading again to the formation of Frenkel pairs. Apart from the small effects on Jc, systematic increases of the normal-state resistivity were found, which fit well into the picture of radiation-induced disorder and its consequences for charge-carrier scattering. The next example will address the improvement of critical current densities and irreversibility lines by statistically distributed spherical defects, such as those introduced by fast neutron irradiation. The basic principles have already been discussed in sect. 6.3, in particular also the complicated interactions between the radiation-induced point defects (To depression, reduction of material anisotropy due to isotropically distributed scattering centres) and the as-grown defect structure, flux pinning by small defect clusters and by large cascades, as well as the role of defect mobility at ambient temperature or upon annealing at elevated temperatures (Sauerzopf et al. 1995, Sauerzopf 1998). Some of the salient features are summarised in fig. 35, where Jc~b is plotted versus fast-neutron fluence for two Y-123 single crystals. The crosses refer to sequential irradiation without intermediate annealing, i.e., to the ftux-pimling action of both the small clusters and the larger collision cascades, whereas the squares show the same irradiation treatment, but interrupted by annealing cycles at 250°C after each irradiation step. Starting from very low Jc's in these single crystals, both kinds of defects initially enhance Jc, but the strongly pinning cascades take over after reaching a significant concentration (~2 × 1022 m3). Note also that Jc varies linearly with cascade concentration. The increase of critical current densities in single crystals is generally high (by factors on the order of 100), and the values of Jc can reach those of thin films, as shown in fig. 36. Due to the statistical arrangement of the newly introduced defects, enhancements of Jc are also observed for other orientations of the magnetic field, in particular of course for H II a,b. In this case, magnetic measurements only allow one to deduce the critical current density J~. The relative contributions of smaller and larger defects are depicted in fig. 37. Less dramatic effects are found in melt-textured bulk materials, but even factors

232

H.W. WEBER

I\

'

I

'

I

'

I

I

'

,o !+%

I

,,Boo

_----.+

1 0:

'

"°~ e'~e<..~+..+~

T=7 7 K

!

Hfie

i

-"~%~.+~+ \

o

L --u--unirradiated 7

10 t

~

\

__O__1X1021 r~ 2

i ~

I -A-8x1021 m 2

" ~

r -+-thin 1 0 6kl '

0

film, unirradiated I , I

1

2

,

~

4-

7x

X

X X I\

3

,

4-... I\

4

,

I

5

laoH (T) Fig. 36. Comparison of critical cm'rent densities in Y-123 thin films and single crystals, in the latter case at various fast neutron fluences.

on the order of 10 can be considered to be of significant technological importance in this case. First results on a Nd-123 monolith are shown in fig. 38 (Kern 1997, Weber 1996), which demonstrate the feasibility of Jc's well above 109 A m -2 at low fields and still above 108 A m -2 at 6 T. Although the irreversibility lines are rarely found to change upon irradiation in melt-textured materials (due to the presence of strongly pinning defects in the pristine samples), some small effects still occur in Nd-123, leading to an irreversibility field of ~11 T at 77 K (H 11c), very close to the best result reported so far for melt-textured specimens exposed to high-pressure oxidation (Wolf et al. 1997). Only the exposure of R-123 thin films to fast neutron irradiation has generally led to negative results (Schalk et al. 1992b), which has been attributed to the optimised preirradiation defect configurations that are being formed during the growth process. In fact, irradiation to various fluences has always left the critical currents nearly unchanged except for those transport measurements in which both the field and the current were oriented in the a,b planes (but still perpendicular to each other). In this case (fig. 39, Schalk 1992), Jc drastically decreases (for H parallel to the planes), which is taken as the signature of disorder-induced weakening of the intrinsic pinning effect. Melt-textured materials (Y-123, Nd-123) with U admixtures represent a highly promising system for high-J~ applications due to the possibility of forming randomly oriented fission tracks (cf. sect. 6.3). Results on the critical current densities (taken by the flux-profile technique, i.e. in the nearly unrelaxed state) are shown in fig. 40 (Weinstein et al. 1998, Eisterer et al. 1998). Jc following thermal neutron irradiation to the optimal

FLUX PINNING

233

6 E < % ~. 4

~

2 Fig. 37. Critical current density along the crystallographic c-direction (/4 II a,b) of an Y-123 single crystal as function of fast neutron fluence. Open symbols refer to results after anne~fling (Sauerzopf 1998).

0 0 1 2 3 4 neutron fluence (102~ m2, E>O.1MeV)

16

14t 0-%

10£

Nd-123

~°~O~,o~

~

. . . . irradiated • 2"1021m-2

I~,\

12

\

--.-- unirr. --o_ 2xlO~m2

~o~ =2

o

~o

10~

~-~

(a) 0

1

2

21 (b)

"~.

3 4 poH (T)

5

6 Temperature I K)

Fig. 38. Fast neutron irradiation of a Nd-123 melt-textured superconductor (Kern 1997): (a) critical current density (H II e) versus field at 77 K (prior to and following irradiation to a fluence of 2 × 10alm -2, E > 0.1 MeV); (b) irreversibility lines for H II c, same conditions as above.

8.0x108

A

6.0x10 e

.,---

E

4.0x10 ~

• unirr

•

irr

m

2.0x10 {

0.0 |

-20

i

|

0

m

I

20

i

i

40

m

I

60

m

|

80

I

100

e(o) Fig. 39. Fast neutron irradiation of Y- 123 thin films, H II a,b: 0 = 0 (Schalk 1992).

234

H.W. WEBER i

!

.

i

,

J ,

i

,

,

(a)

,

i

E

<

1o~

~°~o

-o o

&

o

10~ I ~ Z ~ zx~,.. \ A ~

f

0

2

4

.~

\

m--m~tL ~ ~ ~

i

77K

o-~-~_O___o

,

i

,

(b)

D

unirr.

• x

1.8-102°m-2 4.2"102°m~2

10~° s ~ ,

'

,

'DC

<

•ZXD 50K77K77K uNrr[rr'Unirr

6

8

I%H (Y)

•

50K irr

•

30K irr

10

12

10z 14

0

1

2

3

4

5

6

7

#oH (T)

Fig. 40. (a) Critical current densities (H IIc) in melt-textured Y-123 with an admixture of 0.3 wt% uranium at various temperatures. The open symbols refer to a reference sample, the solid symbols to a sample exposed to a thermal neutron fluence of 4x 1020m-2 (Weinstein et al. 1998). (b) Sequential thermal neutron irradiation of a melt-textured Y-123 superconductorwith an admixture of 0.3 wt% uranium. Critical current densities (H IIc) versus field at 77 K (open squares: unirradiated, solid squares: 1.8x 102urn 2, crosses: 4.2× 102°m 2, Eisterer et al. 1998). fluence (4× 1020 m 2) for this particular U content (0.3 wt%) is shown in fig. 40a at three temperatures and compared to typical data prior to irradiation. The resulting Jc'S are superior to any previously reported data. More systematic work has started only recently; the evolution o f Jc at 77 K with thermal neutron fluence for a material with a U content of 0.3 wt% is shown in fig. 40b. With the next example I will return to more fundamentally oriented studies, although the results on Jo are spectacular in some cases. Exposure o f Y-123 single crystals to heavy-ion irradiation has led to a wealth o f information, which can be briefly summarised as follows. First, the critical current densities are enormously enhanced, particularly at high temperatures where the as-grown defects lose their pinning capability. An example referring to 5 8 0 M e V Sn ion irradiation at an angle o f 2 ° with respect to the crystallographic c-axis is shown in fig. 41 (Civale et al. 1991). The ion fluence is expressed (rather inconveniently) by the so-called "matching field" B . , i.e., the field at which the density o f flux lines matches the density o f the induced columnar tracks. (Since each incoming ion creates one such defect, a ttuence of, say, 3 x 1015 m -2 corresponds to B , - 6 T). The results o f fig. 41 confirm the huge increase of Jc, particularly at 77 K (note the open squares near Jc = 0 in the unirradiated state). The dotted line indicates the effect of proton irradiation on a similar crystal (Civale et al. 1990), which is characteristic of the flux-pinning action o f small point-defect clusters. Clearly, the additional number of small defects also contributes to Jo, but the irreversibility line remains unchanged. On the contrary, upward shifts o f the IL occur upon heavy-ion irradiation (fig. 42, Civale and Krusin-Elbaum 1997) with no hint o f saturation, indicating the presence of flux-

FLUX PINNING

1,2- ~

Y = 5 K

(a)

H II defects//c-axis

.......

<

235

0.8-

=

0.4-

s ~ t

<

T = 77 K ~--

tt

i 0

(b) ~

%

•

B4'=0 T

• B.=3,

i

•

2

| 4

go H (T)

•

Fig. 41. Critical current densities (Hllc) as a function of field in an Y-123 single crystal irradiated with 580MeV Sn ions to various fluences (expressed by the matching fields B.): (a) 5K, (b) 77K. For comparison, the largest Jo's obtained by proton irradiation of a similar crystal are shown by the dotted lines (Civale 1997).

pinning centres with strongly enhanced pinning potentials, in agreement with the first observation of this kind reported by Kritscha et al. (1990b) for fast-neutron-irradiated Bi-2212 single crystals. However, care must be taken at higher fiuences, since some deterioration of the superconducting matrix, as indicated, e.g., by a Tc depression, cannot be avoided and the amount of non-superconducting material will become of increasing significance (recall that a single columnar track has a diameter of ~10nm and extends through the entire thickness of the crystal, i.e. 10 to a maximum of 100 gm). Finally, irradiation experiments on single crystals, thin films and multilayers should be mentioned, where the incident beam strikes the sample at a certain angle with respect to the main crystallographic directions (Civale et al. 1991, 1992, Hardy et al. 1996, Holzapfel et al. 1993, Kraus et al. 1994b, Schmitt et al. 1991). First, single crystals irradiated at an angle of 30 ° from the c-axis (580 MeV Sn ions) and magnetised at angles of +30 ° or - 3 0 ° from the c-axis show an enormous difference in flux pinning at high temperatures (fig. 43a), since the flux lines are aligned parallel to the tracks in one case and cross the tracks under an angle of 60 ° in the other case. This large unidirectional anisotropy nicely confirms the existence of a 3D flux-line lattice at high temperatures. In addition, the significant differences in the hysteresis loops at high fields demonstrate the role of the defect - fluxline lattice orientation for the pinning potential and the irreversibility line. The fact that the enhancements Of Jc are nearly the same for both orientations at low temperatures, is in agreement with many other results and confirms the role of flux pinning by the as-grown defect structure at these temperatures. Under more general geometrical conditions (e.g., Hardy et al. 1996) or upon rotating the single crystal with respect to the magnetic field direction (e.g., Kolesnik et al. 1997), a "lock-in" of the flux line at the defect orientation

236

H.W. WEBER '

I

...

i

2 B.=0.6T ' v ~ v ~

0

•~

~tt

,

I

0.90

0,95

1.00

t=T/T c

Fig. 42. Shift of the irreversibility line (H II c) in an Y-123 single crystal following 1.08 GeV Au ion irradiation to various fluences (Civale and Krusin-Elbaum 1997).

or the a,b planes (Feinberg and Villard 1990, Feinberg 1992) and/or the formation of kinks and "stair-case" patterns had to be invoked. Similar conclusions were drawn from transport measurements on thin fihns, which were irradiated parallel to the c-direction and under angles of 30 ° and 60 ° from the c-axis (fig. 43b). The angular dependence of Jo shows in all cases a pronounced maximum for H parallel to the tracks, which can even exceed the intrinsic pinning maximum for H [I a,b. However, if an Y/Pr-123 multilayer is exposed to the same treatment (fig. 43c), the correlation with the track direction is lost, the intrinsic pinning peak reduced (in agreement with the situation depicted in fig. 39) and Jc otherwise generally enhanced. This can only be understood if the 3D flux-line lattice is replaced by a 2D pancake structure (due to the intermittent normal-conducting Pr-123 layers), which are pinned and depinned independently (in the same way as in other 2D materials such as Bi-2212) and, therefore, do not correlate with preferential defect orientations. To conclude this section, I wish to just show - without comments - the reward of many years of work, which should allow us to proceed with applications of Y-123 materials in all those areas that require conductors in the form of tapes or wires, i.e., the successful development of coated conductors. In view of all the favourable properties of this class of high-temperature superconductors, i.e., their high irreversibility line, their correlated fluxline lattice, the flux-pinning properties, etc., which all manifest themselves at their best in Y-123 thin films, the fabrication of thick films on technical substrates without significant weak link problems is to be considered as a major breakthrough in the field (e.g., Norton et al. 1996, Freyhardt et al. 1997). Figure 44 demonstrates that the best Jo-B characteristics achieved so far for Y-123 thin films are nearly matched by the coated conductors and thus provide us with means of producing long lengths (although at present over unacceptably

FLUX PINNING

i

,

i

~;

i

I

i

,

~

237

i

~

~

~

o.~

~

o

~o. •~ ' ~

~

,~ ~ -~ ';

(~m/Vo~Ol) ~r

o~.

io

~ - ~ " .~

~

e~

~.~ r~

(m/V 90I) IN

( m / V ~0I) I ~

"~"

~Z

~

o ~

c~

238

H.W WEBER

10"

,,,..4~_-.~1.1p.m YBCO on SrTiO 3

109 ~ lOS

'

-

~

tl a,b

H

on rolled Ni

"~

~"k,~

107~-~ B''2223/Agta~ ~'~x ,

106 0

\.~J-- TI- 1223/YSZ tape \

. . . 1 2

. . . 3 4

. 5

"% "N HIIc

. 6

7

~on (T)

Fig. 44. Transport critical current densities in a coated conductor (1.4 g m Y-123 on rolled Ni) compared to results on Y-123 thin films on SrTiO3, Ag-sheathed Bi-2223 tapes and TI-1223 on YSZ tapes (Norton et al. 1996).

long periods of time) of high-Jc material on planar or curved substrates. Progress in this field has been so rapid that any further discussion would be outdated by the time this article goes into print.

8. Field mapping and levitation forces

Field mapping of the remnant magnetisation and measurements of the levitation forces have established themselves as rather independent topics with strong emphasis on application-oriented aspects. These experiments are performed exclusively on melttextured superconductors and are aimed at optimising the trapped field. This can obviously be achieved by the highest possible Jo, but requires at the same time the largest possible sample size [because Mirr(H = 0) is directly proportional to this parameter]. Hence, there is not much to add in the context of this article, since the ways and possibilities of enhancing Jo have been outlined in the preceding sections and the technical issues related to the fabrication of large "single-grain" pellets are addressed in other chapters of this book. Nevertheless, I wish to add a few comments, in particular on the experimental aspects of the field-mapping procedures, which are sometimes treated rather casually in the literature. An early overview of the subject can be found in the monograph by Murakami (1992, chapter 14); fundamental general aspects have been discussed by Brandt (1989); and more recent developments have been compiled in conference proceedings, e.g., EUCAS '95 (Dew-Hughes 1995), Low Temperature Engineering and Cryogenics Conference '96 (Scurlock 1997), or ASC '97 (Clark 1997). The general task is to magnetise the sample to at least twice the full Bean penetration field (to ensure uniform and maximal flux density gradients after demagnetisation), to reduce the external field to zero and to measure either the distribution of the trapped magnetic induction or the force exerted on some permanent magnet. Each of these steps can be done in different ways, which should in principle lead to identical results, but can be subject to subtle external effects. The magnetisation of the sample ("activation") can either be done by field cooling to the desired temperature (in general 77 K) or by

FLUX PINNING

239

raising the field at this temperature. The latter may have some advantages, since largebore superconducting magnets can be avoided and pulsed devices used instead. However, in this case, a detailed analysis of the pulse time and the required magnitude of the field are essential (Gruss et al. 1999, Fujimoto et al. 1997, Morikawa and Tsuchimoto 1997), in order to achieve the fully penetrated state. Furthermore, cracking under the enormous forces represents a major problem that was shown to not only depend on the static parameters, but also on the dynamics of the process (Ren et al. 1995, Weinstein et al. 1996, Fuchs et al. 1996). This may actually represent the most serious limitation for further progress in the field. Once the remnant state is established, field mapping is usually done by a scanning Hall probe device on an x - y sledge, which can be computer-controlled for automatic data acquisition (e.g. Nagashima et al. 1997). A typical example is shown in fig. 45 (Novak 1998), which demonstrates a uniform Bean-like field profile over a pellet with a diameter of 25 m m and a thickness of 13.6 m m (fig. 45a). This confirms in a qualitative manner that the supercurrent flow is everywhere unimpeded by weak links, that the activation had been done correctly into the fully penetrated state, and that the application of the Bean model will be well justified. For a quantitative analysis of the maximum trapped field Bmax and the associated critical current densities, numerical calculations require due consideration of a number of issues, which are only rarely mentioned in the literature. First, of course, the exact geometry of the system has to be used in the calculations (plate versus disk). Second, the properties of the Hall probe device have to be considered (the size and the shape of the probe will determine a certain amount of spatial averaging of the profile, which could mask small-scale inhomogeneities of the material). Third, the thickness of the Hall probe and its coating or substrate will determine the size of the gap between the surface of the superconductor and the measuring device (the same problem will also be caused by non-planar surfaces of the pellet or a variable sample thickness over the diameter). And finally, the time scale of the experiment has to be specified because of thermal activation and the corresponding relaxation of the profile. As an example, the influence of the gap size is demonstrated in fig. 45b. This suggests that the Bmax "extrapolated" to a gap size of zero would be most adequate. Furthermore, an explicit determination of the time dependence of Bm~x would be highly desirable. Although this is often quoted in the literature to n~cely follow a logarithmic law and to be approximately 5% per decade (e.g., Weinstein et al. 1996, see also fig. 45c), it could be worthwhile to use a certain "prescription", say t = 1.000 s, for specifying B . . . . Finally, in particular when addressing large thick pellets, it can be very useful to check the sample on both sides, since cracks or other major imperfections may not go right through the entire material thickness (cf. fig. 45d, which refers to the "bottom" side of an otherwise "normal" sample). A consistency check of the above can be made by measuring the magnetisation of the sample by SQUID, vibrating sample magnetometry or other techniques, which is also sometimes used to assess Jo as an input parameter for the above numerical calculations. Good agreement is found on small samples. However, this can only be of limited relevance at present, since the pellet sizes of interest, at up to 50 mm diameter, by far exceed the spatial capabilities of the magnetometers.

240

H.W. WEBER

10

O5

O0 15

-0 5 o

-10 position

25O

on center

line (ram)

-

.... 230-

(c) "-..........

220 210 -

g

200 -

~'.. "-.

N

; 8(, -

"-~(.

170 -

~-....~ l-~il:i!t '

time

(s)

q~

,j.poa~t, . . . .

Fig. 45. Field mapping of the remnant magnetisafion in large Y-123 pellets (Novak 1998): (a) flux profile at 77 K in a pellet with a diameter of 25 mm and a thickness of 13.6 mm, activated at 77 K and 2 T (gap: 0.5 ram); (b) influence of the gap size on the trapped field in a rectangulm" platelet with dimensions 10x 10× 1 mm3; (c) influence of the relaxation at 77 K on the maximum trapped field (activation at 77 K in a field of 2 T); (d) flux profile showing the influence of a crack on the bottom of a pellet with a diameter of 25 mm and a thickness of 13.6ram, activated at 77K and 2T (gap: 0.6mm).

Further recent work refers to the analysis of trapped fields when several pellets are stacked together (e.g. Nagashima et al. 1997), to possibilities of reducing flux creep by special procedures (Liu et al. 1995, Weinstein et al. 1996, Levin et al. 1998), and to enhancements of Jo by radiation techniques (Weinstein et al. 1996, 1998, Eisterer et al. 1998, Novak 1998). Successful reductions of the creep rate were achieved by postactivation cooling by several degrees or by cooling the sample still during activation. Record trapped fields of 10.1 T at 4 2 K could be achieved in this way and maintained nearly unchanged for more than three hours (Weinstein et al. 1996). Alternatively, heat pulses were reported to reduce creep in (Y,Pr)-123 single crystals (Levin et al. 1998). Enhancements of Jc by statistically distributed fission tracks (Weinstein et al. 1998) led to an increase of Bmax by factors of ~5 at 7 7 K in Y-123 pellets of 20mm diameter and 8 mm thickness. Similar results were obtained upon fast neutron irradiation of Y- and Sm-123 plates (10× 10x 1 mm3), cf. fig. 46.

FLUX PINNING

150

800

~ ~5

241

4C 2q

10

-10

Fig. 46. Trapped-field distribution at 77 K in a rectangular platelet with dimensions 10 x 10 × 1 I n I n 3 , activated at 77 K in a field of 2 T and extrapolated to gap zero (Novak 1998): (a) trapped-field distribution in the unirradiated s t a t e : Bz, max = 156roT; (b) trapped-field distribution in the irradiated state (fast neutron fluence: 2x 10zl m -2, E>0.1 MeV): B ....... =812mT.

With regard to measurements of the levitation forces, many of the above comments apply in a similar way. Crucial parameters are again Jc, Bmax, the size and thickness of the superconductor, the distance between superconductor and magnet, and, in addition, the magnet material, the size of the magnet, the relative size of the superconductor versus the size of the magnet, etc. For more details, cf., e.g., Murakami (1992), Campbell and Cardwell (1997), Teshima et al. (1997).

9. Summary As stated in the introduction of this article, I have tried to focus my review of published work on flux pinning in R-123 superconductors on those aspects which are in principle suited for achieving high critical current densities in these materials. We should not forget at this stage that - although discovered among the first high-temperature superconducting materials - R-123 superconductors still hold the highest potential for high critical current densities at the boiling temperature of liquid nitrogen among all other members in the family of the cuprates, and are potentially challenged in this respect only by T1-1223 or Hg-1223, if suitable processing technologies can be developed. This unmatched performance at high temperatures is primarily related to the (nearly) threedimensional character of superconductivity, which allows the formation of a 3D flux-line lattice and, consequently, facilitates pinning by a given crystallographic defect structure in comparison to the (more) 2D superconductors forming flux-line segments (pancakes) of comparatively high mobility under a Lorentz force. The main issue that will require close attention, also in the foreseeable future, is the optimisation of the defect structures for flux pinning. The record performance of R-123

242

H.V~ WEBER

thin films can still be understood only partly in terms of the underlying defect structures, but has been approached recently by the fabrication technologies for coated conductors, which are in principle not too different from thin-film growth techniques, and have led to success by avoiding the formation of grain boundaries with too large grain-boundary angles. Obviously, this crystallographic defect has to be avoided, and my best guess about the most likely defect structure in 'the thin films is the formation of a dense array of extended planar defects along the crystallographic c-direction. The growth conditions of other forms of R-123 superconductors are too different to allow a one-to-one projection of these defects in these cases. Important defect structures for flux pinning in single crystals and melt-textured materials seem to be clusters of oxygen-deficient lattice cells, normal-conducting precipitates, and arrays of stacking faults, but their control is rather tedious and in many cases determined by the growth conditions. A remarkable exception is provided by radiation techniques, which allow, under certain conditions, extremely well-controlled modifications of the defect structures for flux pinning, and have led to record Jc'S both in single crystals and in melt-textured materials. The extent to which they may be useful for applications, is still under discussion, and certainly depends on the envisaged application. It is safe to state, in conclusion, that progress with our understanding of issues related to improve flux pinning in R-123 superconductors has been breathtaking over the past decade, and will certainly remain in the focus of further research, particularly with respect to all micro- and nano-structural aspects of material properties and defect configurations.

Acknowledgements I am pleased to acknowledge the continuous, most fruitful and scientifically challenging cooperation with Dr. Franz M. Sauerzopf during the past (many) years, who also proved to be indispensable for the day-to-day operation of and the working atmosphere in a group of up to 25 members. The contributions of many brilliant undergraduate and graduate students as well as of post-doctoral fellows to the research activities described in this article are gratefully acknowledged. Furthermore, I wish to thank many colleagues from abroad, who contributed to our research in various ways, in particular Professor H.C. Freyhardt, G6ttingen, Dr. M. Murakami, Tokyo, Dr. M.A. Kirk and Dr. B.W. Veal, Argonne, Professors R. Weinstein and K. Salama, Houston, Professor D. Bfiuerle, Linz, Professor R. Griessen, Amsterdam, Professor R Kes, Leiden, and Professor Y. Yeshurun, Ramat-Gan. I also wish to thank many colleagues for their willingness to provide me with original data or graphic files of some of their figures reproduced in this article. In conclusion, I wish to thank the Austrian Science Foundation, Vienna, for fimding more than ten research projects on HTS, and the European Union for funding two Brite Euram projects and two networks on the subject (HCM and TMR).

FLUX pINNING

243

References Abrikosov, A.A., 1957, Soy. Phys. JETP 5, 1174. Abulafia, Y., A. Shaulov, Y. Wolfus, R. Prozorov, L. Burlachkov and Y. Yeshurun, 1995, Phys. Rev. B 75, 2404. Abulafia, Y., A. Shaulov, Y. Wolfus, R. Prozorov, L. Bttrlachkov, Y. Yeshurtm, D. Majer, E. Zeldov, H. Wtihl, VB. Geshkenbein and VM. Vinokur, 1996, Phys. Rev. Lett. 77, 1596. Adaktylos, I., and H.W Weber, 1977, Philos. Mag. 35, 983. Adaktylos, I., E. Schachinger and H.W Weber, 1977, J. Low Temp. Phys. 26, 533. Alers, RB., 1957, Phys. Rev. 105, 104. Almasan, C.C., M.C. de Andrade, Y. Dalichaouch, J.J. Neumeier, C.L. Seaman, M.B. Maple, R.R Guertin, M.V Kuric and J.C. Garland, 1992, Phys. Rev. Lett. 69, 3812. Anderson, RW., 1962, Phys. Rev. Lett. 9, 309. Antesberger, G., and H. Ullmaier, 1974, Philos. Mag. 29, 1101. Bean, C.R, 1962, Phys. Rev. Lett. 8, 250. Bednorz, LG., and K.A. Mfiller, 1986, Z. Phys. B 64, 189. Biersack, J.R, and L.G. Haggmark, 1980, Nucl. Instrum. Methods 174, 257. Blatter, G., M.V Feigel'man, VB. Geshkenbein, A.I. Larkin and V.M. Vinokur, 1994, Rev. Mod. Phys. 66, 1125. B6hmer, C., G. Brandstfitter and H.W. Weber, 1997, Supercond. Sci. Technol. 10, A1. Brandst/itter, G., EM. Sanerzopf, H.W Weber and B.W Veal, 1996, Czech. J. Phys. 46(S3), 1681. Brandst/itter, G., EM. Sauerzopf and H.W. Weber, 1997, Phys. Rev. B 55, 11693. Brandstfitter, G., EM. Sauerzopf, H.W. Weber and B.W. Veal, 1998a, in: Critical Currents in Superconductors for Practical Applications, eds L. Zhou, H.W. Weber, E.W Collings (World Scientific, Singapore) p. 325. Brandstfitter, G., M.C. Frischherz, L. Kratzwald, H.W. Weber and A.E. Petrov, 1998b, Physica C 307, 197. Brandt, E.H., 1989, Science 243, 349. Brandt, E.H., 1995, Rep. Prog. Phys. 58, 1465. Brandt, E.H., 1996, Phys. Rev. B 54, 4246. Breit, V, R Schweiss, R. Hanff, H. Wiihl, H. Claus, H. Rietschel, A. Erb and G. Mfiller-Vogt, 1995, Phys. Rev. B 52, R15727. Brezgunov, M.M., S.V Belogurov, A.E. Petrov, A.N. Karpov, I.V, Kudrenitskis, VN. Kovalev,

E.N. Kurkin, A.A. Makletsov, Yu.G. Morozov and M.D. Nersesyan, 1990, Superconductivity 3, 843. Bulaevskii, L,N., M. Ledvij and VG. Kogan, 1992, Phys. Rev. Lett. 68, 3773. Campbell, A.M., 1969, J. Phys. C 2, 1492. Campbell, A.M., and D.A. Cardwell, 1997, Cryogenics 37, 567. Campbell, A.M., and J.E. Evetts, 1972, Critical Currents in Superconductors (Taylor & Francis, London). Cava, R.J., A.W. ttewat, E.A. Hewat, B. Batlogg, M. Marezio, K.M. Rabe, J.J. Krajewski, W.E Peck Jr and L.W. Rupp Jr, 1990, Physica C 165, 419. Chien, T.R., W.R. Datars, B.W Veal, A.P. Paulikas, R Kostid, (2. Gu and Y. Jiang, 1994, Physica C 229, 273. Chikumoto, N., and M. Mm'akami, 1996, in: Proc. 10th Anniversary HTS Workshop on Physics, Materials and Applications, eds B. Batlogg, C.W. Chu, WK. Chu, D.U. Gubser and K.A. Mtiller (World Scientific, Singapore) pp. 139-142. Civale, L., 1997, Supercond. Sci. Technol. 10, A11. Civale, L., and L. Krusin-Elbaum, 1997, Phys. Rev. Lett. 78, 1829. Civale, L., A.D. Marwick, M.W. McElfresh, T.K. Worthington, A.E Malozemoff, E Holtzberg, J.R. Thompson and M.A. Kirk, 1990, Phys. Rev. Lett. 65, 1164. Civale, L., A.D. Marwick, T.K. Worthington, M.A. Kirk, J.R. Thompson, L. Krusin-Elbaum, Y.R. Sun, J.R. Clem and E Holtzberg, 1991, Phys. Rev. Lett. 67, 648. Civale, L., T.K. Worthington, L. Krusin-Elbaum, A.D. Marwick, E Holtzberg, J.R. Thompson, M.A. Kirk and R. Wheeler, 1992, J. Met. 44, 60. Civale, L., L. Krusin-Elbaum, J.R. Thompson and E Holtzberg, 1994a, Phys. Rev. B 50, 7188. Civale, L., L. Krusin-Elbaum, J.R. Thompson, R. Wheeler, A.D. Marwick, M.A. Kirk, Y.R. Sun, E Holtzberg and C. Feild, 1994b, Phys. Rev. B 50, 4102. Clark, A.E, ed., 1997, IEEE Trans. Appl. Supercond. 7. Clem, J.R., 1991, Phys. Rev. B 43, 7837. Clem, J.R., 1992, in: Magnetic Susceptibility of Superconductors and Other Spin Systems, eds R.A. Hein, T.L. Francavilla and D.H. Liebenberg (Plenum, New York) p. 177.

244

H.W. WEBER

Clem, J.R., and M.W Coffey, 1990, Phys. Rev. B 42, 6209. Cohen, L.E, J.R. Laverty, G.K. Perkins, A.D. Caplin and W. Assmus, 1993, Cryogenics 33, 352. Coote, R.I., J.E. Evetts and A.M. Campbell, 1972, Can. J. Phys. 50, 421. Cribier, D., B. Jacrot, L.M. Rao and B. Farnoux, 1964, Phys. Lett. 9, 106. Cubitt, R., E.M. Forgan, G. Yang, S.L. Lee, D.McK. Paul, H.A. Mook, M. Yethiraj, RH. Kes, T.W. Li, A.A. Menovsky, Z. Tarnawski and K. Mortensen, 1993, Nature 365, 407. D/iumling, M., J.M. Seuntjens and D.C. Larbalestier, 1990, Nature 346, 332. Dfiumling, M., A. Erb, E. Walker, J.Y. Genoud and R. Fliikiger, 1996, Physica C 257, 371. de Groot, EA.J., Y. Zhu-An, R. Yanru and S. Smith, 1991, Physica C 185, 2471. Dew-Hughes, D., 1971, Rep. Prog. Phys. 34, 821. Dew-Hughes, D., 1974, Philos. Mag. 30, 293. Dew-Hughes, D., ed., 1995, Applied Superconductivity 1995, IOP Conf. Set. 148. Dew-Hughes, D., and M.J. Witcomb, 1972, Philos. Mag. 26, 73. Dolan, G.J., E Holtzberg, C. Feild and T.R. Dinger, 1989, Phys. Rev. Lett. 62, 2184. Doyle, R.A., B. Khaykovich, M. Konezykowski, E. Zeldov, N. Morozov, D. Majer, RH. Kes and V Vinokur, 1997, Physica C 282-287, 323. Egi, T., J.G. Wen, K. Kuroda, H. Unoki and N. Koshizuka, 1995, Appl. Phys. Lett. 67, 2406. Eisterer, M., S. T6nies, W. Novak, H.W. Weber, R. Weinstein and R.E Sawh, 1998, Supercond. Sci. Technol. 11, 1101. Erb, A., E. Walker and R. Fliikiger, 1995, Physica C 245, 245. Erb, A., E. Walker and R. Fliikiger, 1996a, Physica C 258, 9. Erb, A., J.Y. Genoud, E Marti, M. D/iumling, E. Walker and R. Fliikiger, 1996b, J. Low Temp. Phys. 105, 1023. Erb, A., E. Walker, J.Y. Genoud and R. Flfikiger, 1997, Physica C 282-287, 2t45. Evetts, J.A., and J.M.A. Wade, 1970, J. Phys. Chem. Solids 31,973. Feenstra, R., D.K. Christen, C.E. Ktabunde and J.D. Budai, 1992, Phys. Rev. B 45, 7555. Feigel'man, M.V, V.B. Geshkenbein, A.I. Larkin and V.M. Vinokur, 1989, Phys. Re,/. Lett. 63, 2301. Feigel'man, M.V., VB. Geshkenbein and VM. Vinokur, 1991, Phys. Rev. B 43, 6263.

Feinberg, D., 1992, Physica C 194, 126. Feinberg, D., and C. Viltard, 1990, Phys. Rev. Lett. 65, 919. Fleischer, R.L., J.R. Hart Jr, K.W Lay and EE. Luborsky, 1989, Phys. Rev. B 40, 2163. Fossheim, K., E.D. Tuset, T.W Ebbesen, M.MJ. Treacy and J. Schwartz, 1995, Physica C 248, 195. Freyhardt, H.C., 1971a, Philos. Mag. 23, 369. Freyhardt, H.C., 1971b, Philos. Mag. 23, 345. Freyhardt, H.C., J. Hoffmann, J. Wiesmann, J. Dzick, K. Heinemann, A. Isaev, E Garcia-Moreno, S. Sievers and A. Usoskin, 1997, IEEE Trans. AppL Supercond. 7, 1426. Frischherz, M.C., M.A. Kirk, J.P. Zhang and H.W. Weber, 1993, Philos. Mag. A 67, 1347. Frischherz, M.C., M.A. Kirk, J. Farmer, L.R. Greenwood and H.W Weber, t994, Physica C 232, 309. Frischherz, M.C., EM. Sauerzopf, H.W. Weber, M. Murakami and G.A. Emel'chenko, 1995, Supercond. Sci. Teclmol. 8, 485. Fuchs, G., G. Krabbes, R Schfitzle, R Stoye, T. Staiger and K.-H. Mfiller, 1996, Physica C 268, 115. Fujimoto, H., K. Taguchi, M. Murakami, N. Nakamura and N. Koshizuka, 1992, Cryogenics 32, 954. Fujimoto, H., K. Kawann, T. Higuchi, K. Nagashima and M. Murakami, 1997, Physica C 282-287, 2327. Gammel, RL., D.J. Bishop, J.R Rice and D.M. Ginsberg, 1992, Phys. Rev. Lett. 22, 3343. Gerber, C., D. Anselmetti, J.G. Bednorz, J. Mannhart and D.G. Schlom, 1991, Nature 350, 279. Gerhiiuser, W, G. Ries, H.W. Neumfiller, V~ Schmidt, O. Eibl, G. Saemann-lschenko and S. Klaumfinzer, 1992, Phys. Rev. Lett. 68, 879. Geshkenbein, VB., A.I. Larkin, M.V. Feigel'man and VM. Vinokur, 1989, Physica C 162 164, 239. Geshkenbein, V.B., V.M. Vinokur and R. Fehrenbacher, 1991, Phys. Rev. B 43, 3748. Giapintzakis, J., M.A. Kirk, W.C. Lee, J.E Rice, D.M. Ginsberg, I.M. Robertson and R.L. Wheeler, 1992, in: Layered Superconductors: Fabrication, Properties and Applications, MRS Syrup. Proc. 275, eds D.T. Shaw, C.C. Tsuei, T.R. Schneider and Y. Shinohara (MRS, Pittsburgh) pp. 741746. Giapintzakis, J., D.M. Ginsberg, M.A. Kirk and S. Ockers, 1994a, Phys. Rev. B 50, 15967. Giapintzakis, J., R.L. Neiman, D.M. Ginsberg and M.A. Kirk, 1994b, Phys. Rev. B 50, 16001. Ginzburg, VL., and L.D. Landau, 1950, Zh. Eksp. Teor. Fiz. 20, 1064. G6m6ry, E, 1997, Supercond. Sci. Technol. 10, 523.

FLUX PINNING Gordeev, S.N., W. Jahn, A.A. Zhnkov, H. Kfipfer and T. Wolf, 1994, Phys. Rev. B 49, 21. Gray, K.E., D.H. Kim, B.W. Veal, G.T. Seidler, T.E Rosenbanm and D.E. Farrell, 1992, Phys. Rev. B 45, 10071. Griessen, R.R, M.J.G. Lee and D.J. Stanley, 1977, Phys. Rev. B 16, 4385. Gruss, S., G. Fuchs, G. Krabbes, R Sch/itzle, J. Fink, K.-H. Miiller mad L. Schultz, 1999, IEEE Trans. Appl. Supercond. 9, 2070. Gu, H., C. Colliex, S. Senoussi, C. Aguillon-LeviUain and P. Manuel, 1993, Philos. Mag. A 68, 19. Gyorgy, E.M., R.B. van Dover, K.A. Jackson, L.E Schneemeyer and J.V Waszczak, 1989, Appl. Phys. Lett. 55, 283. Hardy, V., D. Groult, M. Hervieu, J. Provost, B. Raveau and S. Bouffard, 1991, Nucl. Instrum. Methods B 54, 472. Hardy, V, D. Groult, J. Provost and B. Ravean, 1992, Physica C 190, 289. Hardy, V, A. Wahl, S. Hebert, A. Ruyter, J. Provost, D. Groult and C. Simon, 1996, Phys. Rev. B 54, 656. High-Tc Update, 1987 2000, published bi-weekly by the US DOE, Ames Laboratory, USA, see also htt-p:/ /www.iitap.iastate.edu/htcu/hicu.html. Hillmarm, H., and D. Hanck, 1972, in: Proc. Applied Superconductivity Conf. (Annapolis), eds H.M. Long and W.E Gauster (IEEE, New York) p. 429. Hirth, J.P., and J. Lothe, 1982, Theory of Dislocations (Wiley, New York) ch. 15. Holzapfel, B., G. Kreiselmeyer, M. Kraus, G. Saemann-Ischenko, S. Bouffard, S. Klanmtinzer and L. Schultz, 1993, Phys. Rev. B 48, 600. Huebener, R.P., R.T. Kampwirth and VA. Rowe, 1972, Cryogenics 12, 100. Jin, S., T.H. Tiefel, R.C. Sherwood, M.E. Davis, R.B. van Dover, G.W. Kammlott, R.A. Fastnacht and H.D. Keith, 1988, Appl. Phys. Lett. 52, 2074. Jones, E.C., D.K. Christen, J.R. Thompson, R. Feenstra, S. Zhu, D.H. Lowndes, J.M. Philips, M.R Siegal and J.D. Budai, 1993, Phys. Rev. B 47, 8986. Jorgensen, J.D., H. Shaked, D.G. Hinks, B. Dabrowski, B.V~ Veal, A.P. Paulikas, LJ. Nowicki, G.W. Crabtree, W.K. Kwok, L.H. Ntmez and H. Claus, 1988, Physica C 153 155, 578. Jorgensen, J.D., B.W Veal, A.P. Paulikas, L.J. Nowicki, G.V~ Crabtree, H. Claus and W.K. Kwok, 1990, Phys. Rev. B 41, 1863.

245

Kern, C., 1997, Diploma Thesis (Technical University of Vienna). Kes, EH., J. Aarts, J. van den Berg, C.J. van der Beek and J.A. Mydosh, 1989, Supercond. Sci. Technol. 1, 242. Kes, RH., J. Aarts, V.M. Vinokur and CJ. van der Beek, 1990, Phys. Rev. Lett. 64, 1063. Kirchner, H., 1968, Phys. Lett. 26A, 651. Koblischka, M.R., and R.J. Wijngaarden, 1995, Supercond. Sci. Technol. 8, 199. Kolesnik, S., H.W. Weber, T. Skoskiewicz, M. Sawicki and W. Sadowski, 1997, Physica C 282-287, 1961. Konczykowski, M., E Rullier-Albenque, E.R. Jacoby, A. Shaulov, Y. Yeshurun and R Lejay, 1991, Phys. Rev. B 44, 7167. Koshelev, A.E., 1994, Phys. Rev. B 50, 506. Kramer, E.J., 1973, J. Appl. Phys. 44, 1360. Kramer, E.J., and C.L. Bauer, 1967, Philos. Mag. 15, 1189. Kraus, M., M. Leghissa and G. Saemann-Ischenko, 1994a, Phys. B1.50, 333. Kraus, M., G. Kreiselmeyer, J. Daniel, M. Leghissa, G. Saemann-Ischenko, B. Holzapfel, E Kummeth, R. Scholz and L.Ya. Vimlikov, 1994b, Nucl. Instrum. Methods B 89, 307. Kritscha, W., 1991, Ph.D. Thesis (Technical University of Vienna). Kritscha, W, F.M. Sauerzopf and H.W. Weber, 1990a, Physica B 165-166, 109. Kritscha, W., EM. Sauerzopf, H.W Weber, G.W. Crabtree, Y.C. Chang and R-Z. Jiang, 1990b, Europhys. Lett. 12, t79. Kritscha, W., EM. Sanerzopf, H.W Weber, G.W. Crabtree, Y.C. Chang and R-Z. Jiang, 1991, Physica C 199, 15. Krusin-Elbaum, L., L. Civale, VM. Vinokur and E Holtzberg, 1992, Phys. Rev. Lett. 69, 2280. Krusin-Elbaum, L., J.R. Thompson, R. Wheeler, A.D. Marwick, C. Li, S. Patel, D.T. Shaw, P. Lisowski and J. Ullmann, 1994, Appl. Phys. Lett. 64, 3331. Ktipfer, H., A.A. Zhukov, A. Will, W. Jahn, R. MeierHirmer, T. Wolf, VI. Voronkova, M. Klgser and K. Saito, 1996, Phys. Rev. B 54, 644. Ktipfer, H., T. Wolf, A.A. Zhukov, C. Lessing, W. Jahn and R. Meier-Hirmer, 1997, Advances in superconductivity, in: New Materials, Critical Currents & Devices, eds R. Pinto, S.K. Malik, A.K. Grover and P. Ayyub (New Age International Limited Publishers, New Delhi) p. 291.

246

H.W. WEBER

Kwok, W.K., U. Welp, G.W. Crabtree, K.G. Vandervoort, R. Hulscher and J.Z. Liu, 1990a, Phys. Rev. Lett. 64, 966. Kwok, W.K., U. Welp, K.G. Vandervoort, Y. Fang and G.W Crabtree, 1990b, Appl. Phys. Lett. 57, 1. Kwok, WK., U. Welp, S. Fleshier, K.G. Vandervoort, G.W. Crabtree, J.Z. Liu, J. Brooks, J. Hettinger and S.T. Harmahs, 1991a, Supercond. Sci. Technol. 4, S106. Kwok, W.K., U. Welp, VM. Vinokur, S. Fleshier, J. Downey and G.W Crabtree, 1991b, Phys. Rev. Lett. 67, 390. Labusch, R., 1969, Cryst. Lattice Defects 1, 1. Larkin, A.I., and Yu.N. Ovchinniknv, 1974, Sov. Phys. JETP 38, 854. Larkin, A.I., and Yu.N. Ovchinnikov, 1979, J. Low Temp. Phys. 34, 409. Lawrence, WE., and S. Doniach, 1971, in: Proc. 12th Int. Conf. on Low Temperature Physics, ed E. Kanda (Keigaku, Tokyo) p. 361. Lensink, J.G., 1993, Ph.D. Thesis (Vrije Universiteit, Amsterdam). Levin, G.A., C.C. Almasan, D.A. Gajewski and M.B. Maple, 1998, Appl. Phys. Lett. 72, 112. Liu, J., R. Weinstein, Y. Ren, R.R Sawh, C. Foster and V Obot, 1995, in: Proc. 1995 Int. Workshop on Superconductivity (Mani), eds S. Tanaka and J.M. Phillips (ISTEC, Tokyo) p. 353. Majer, D., E. Zeldov, H. Shtirkman and M. Konczykowski, 1996, in: Coherence in High Temperature Superconductors, eds G. Deutscher and A. Revcolevschi (World Scientific, Singapore) pp. 271291. Malozemnff, A.R, T.K. Worthington, Y. Yeshurtm, F. Holtzberg and P.H. Kes, 1988, Phys. Rev. B 38, 7203. Mannhart, J., D. Anselmetti, J.G. Bednorz, C. Gerber, K.A. Miiller and D.G. Schlom, 1992, Supercond. Sci. Technol. 5, S125. Martinez, B., E Sandiumenge, N. Vilalta, S. Pifiol, X. Obradors and J. Rabier, 1996, J. Appl. Phys. 80, 5515. McGinn, P., 1995, in: High Temperature Superconducting Materials Science and Engineering, ed D. Shi (Elsevier, Amsterdam) ch. 8, pp. 345 382. Meier-Hirmer, R., H. Kiipfer and H. Scheurer, 1985, Phys. Rev. B 31, 183. Meilikhov, E.Z., and VG. Shapiro, 1991, Superconductivity 4, 1353. Meissner, Vv~, and R. Ochsenfeld, 1933, Namrwissenschaflen 21, 787.

Morikawa, K., and M. Tsuchimoto, 1997, Physica C 282-287, 1863. Mtiller, U., and H.C. Freyhardt, 1996, Philos. Mag. Lett. 73, 63. Murakami, M., ed., 1992, Melt Processed High Temperature Superconductors (World Scientific, Singapore). Murakami, M., K. Yamaguchi, H. Fujimoto, N. Nakamura, T. Taguchi, N. Koshizuka and S. Tanaka, 1992, Cryogenics 32, 930. Murakami, M., S.I. Yoo, T. Higuchi, N. Sakai, J. Weltz, N. Koshizuka and S. Tanaka, 1994, Jpn. J. Appl. Phys. 33, L715. Murakami, M., S.I. Yoo, N. Sakai, K. Sawada, T. Higuchi and N. Chikumoto, 1996, in: Critical Currents in Superconductors, eds T. Matsushita and K. Yamafuji (World Scientific, Singapore) pp. 5762. Nagashima, K., T. Higuchi, J. Sok, S.I. Yoo, H. Fujimoto and M. Murakami, 1997, Cryogenics 37, 557. Norton, D.P., A. Goyal, J.D. Budai, D.K. Christen, D.M. Kroeger, E.D. Specht, Q. He, B. Saffian, M. Paranthaman, C.E. Klabtmde, D.E Lee, B.C. Sales and EA. List, 1996, Science 274, 755. Novak, V~, 1998, Diploma Thesis (Technical University of Vienna). Ossandon, J.G., J.R. Thompson, D.K. Christen, B.C. Sales, H.R. Kerchner, J.O. Thomson, Y.R. Sun, K.W. Lay and J.E. Tkaczyk, 1992a, Phys. Rev. B 45, 12534. Ossandon, J.G., J.R. Thompson, D.K. Christen, B.C. Sales, Y.R. Sun and K.W. Lay, 1992b, Phys. Rev. B 46, 3050. Oussena, M., PAJ. de Groot and S.J. Porter, 1995, Phys. Rev. B 51, 1389. Pearl, J., 1964, Appl. Phys. Lett. 5, 65. Pust, L., J. Kadlecova, M. Jirsa and S. Durcok, 1990, J. Low Temp. Phys. 78, 179. Quafurd, M., K. Heeck, J.G. Lensink, R.J. Wijngaarden and R. Griessen, 1992, Rev. Sci. Instrum. 63, 5526. Rabier, J., P.D. Tall and M.E Denanot, 1993, Philos. Mag. A 67, 1021. Ren, Y., R. Weinstein, J. Liu and R.P Sawh, 1995, Physica C 251, 15. Roas, B., L. Schultz and G. Saemann-Ischenko, 1990, Phys. Rev. Lett. 64, 479. Rollins, R.W., H. Ktipfer and W Gey, 1974, J. Appl. Phys. 45, 5392. Roulin, M., A. Jtmod and E. Walker, 1996, Science 273, 1210.

FLUX PINNING Salama, K., V. Selvamanickam and D.E Lee, 1993, in: Proc. and Properties of High Tc Superconductors, Vol. 1, ed S. Jin (World Scientific, Singapore) ch. 5, pp. 155~11. Samadi Hosseinali, G., H.W. Weber, E. Stangl, S. Proyer and D. Bfiuerle, 1997, Physica C 282-287, 2147. Sandiumenge, E, N. Vilalta, S. Pifiol, B. Martinez and X. Obradors, 1995, Phys. Rev. B 51, 1994. Sandiumenge, E, B. Martlnez and X. Obradors, 1997, Supercond. Sci. Technol. 10, A93. Sauerzopf, EM., 1998, Plays. Rev. B 57, 10959. Sauerzopf, EM., H.R Wiesinger and H.W. Weber, 1990, Cryogenics 30, 650. Sauerzopf, EM., H.E Wiesingel; H.W Weber, G.W Crabtree, M.C. Frischherz and M.A. Kirk, 1992, Supercond. Sci. Technol. 5, S105. Sauerzopf, EM., H.E Wiesinger, H.W. Weber and G.W Crabtree, 1995, Phys. Rev. B 51, 6002. Sauerzopf, EM., M. Werner, H.W Weber, R.A. Suris, V.S. Kharlamov and Yu.V Trushin, 1997, Physica C 282-287, 1333. Schalk, R.M., 1992, Ph.D. Thesis (Technical University of Vienna). Schalk, R.M., H.W. Weber, Z.H. Barber, P. Przyslupsky and J.E. Evetts, 1992a, Physica C 199, 311. Schalk, R.M., H.W. Weber, Z.H. Barber, C.E. Davies, J.E. Evetts, R.E. Somekh and D.H. Kim, 1992b, Supercond. Sci. Technol. 5, $224. Schalk, R.M., G. Samadi Hosseinali, H.W. Weber, S. Proyer, R Schwab, D. B/iuerle and S. Grfindorfer, 1994, Phys. Rev. B 49, 3511. Schalk, R.M., K. Kundzins, H.W Weber, E. Stangl, S. Proyer and D. B/iuerle, 1996, Physica C 257, 341. Schelten, J., H. Ullmaier and W. Schmatz, 1971, Phys. Status Solidi B 48, 619. Schilling, A., R.A. Fisher, N.E. Phillips, U. Welp, D. Dasgupta, WK. Kwok and G.W. Crabtree, 1996, Nature 382, 791. Schmitt, R, E Kummeth, L. Schultz and G. SaemannIschenko, 1991, Phys. Rev. Lett. 67, 267. Sclmack, H.G., 1995, Ph.D. Thesis (Vrije Universeit, Amsterdam). Schnack, H.G., R. Griessen, J.G. Lensink and H.H. Wen, 1993, Phys. Rev. B 48, 13178. Scurloek, A.G., ed., 1997, Cryogenics 37(10). Seidler, G.T., T.E Rosenbaum, D.L. Heinz, LW. Downey, A.P. Paulikas and B.W. Veal, 1991, Physica C 183, 333. Selvamanickam, V, M. Mironova, S. Son and K. Salama, 1993, Physica C 208, 238.

247

Sickafus, K.E., J.O. Willis, RJ. Kung, WB. Wilson, D.M. Parkin, M.P. Maley, EW Clinard Jr, C.J. Salgado, R.P. Dye and K.M. Hubbard, 1992, Phys. Rev. B 46, 11862. Suenaga, M., D.O. Welch and R.C. Budhani, 1992, Supercond. Sci. Technol. 5, S1. Superconductivity Database, 1996, ETL, ISTEC, Japan, http://www.aist.go.jp/RIODB/sprcnd_etl/. Svetchnikov, V.L., VM. Pan, C. Traeholt and H.W Zandbergen, 1997, IEEE Trans. Appl. Supercond. 7, 1396. Tachiki, M., and S. Takahashi, 1989, Solid State Commun. 70, 291. Tesanovic, Z., 1991, Phys. Rev. B 44, 12635. Teshima, H., M. Sawamura, M. Morita and M. Tsucbimoto, 1997, Cryogenics 37, 505. Thompson, J.R., D.K. Christen, Y.R. Sun, H.R. Kerehnet, J.E. Tkaczyk and K.W Lay, 1992, Cryogenics 32, 982. Thompson, J.R., J.G. Ossandon, D.K. Christen, R. Feenstra, B.C. Sales, H.R. Kerchner, E.C. Jones and Y.R. Sun, 1994, in: Critical Currents in Superconductors, ed H.W Weber (World Scientific, Singapore) pp. 70-75. Ting, V~, T. Egi, R. Itti, K. Kuroda, N. Koshizuka and S. Tanaka, 1997, Appl. Phys. Lett. 70, 770. Trfiuble, H., and U. El3mann, 1966, Phys. Status Solidi 18, 813. Trushin, Yu.V, B.J. Ber, VS. Kharlamov and E.E. Zhnrkin, 1996, J. Nucl. Mater. 233-237, 991. Ullmaier, H., 1975, Irreversible Properties of Type II Superconductors (Springer, Berlin). Ullrich, M., D. Mtiller, W Mexner, M. Steins, K. Heinemann and H.C. Freyhardt, 1993, Phys. Rev. B 48, 7513. Umezawa, A., G.W Crabtree, J.Z. Liu, H.W Weber, W.K. Kwok, L.H. Nunez, T.J. Moran, C.H. Sowers and H. Claus, 1987, Phys. Rev. B 36, 7151. van Dalen, A.J.J., 1995, Ph.D. Thesis (Vrije Universiteit, Amsterdam). van Dalen, A.J.J., M.R. Koblischka, R. Griessen, M. Jirsa and G. Ravi Kumar, 1995, Physica C 250, 265. van Dalen, AJ.J., R. Griessen, S. Libbrecht, Y. Bruynseraede and E. Osquiguil, 1996, Phys. Rev. B 54, 1366. Vargas, J.L., and D.C. Larbalestier, 1992, Appt. Phys. Lett. 60, 1741. Vlcek, B.M., M.C. Frischherz, S. Fleshier, U. Welp, J.Z. Liu, J. Downey, K.G. Vandervoort, G.W. Crabtree and M.A. Kirk, 1992, Phys. Rev. B 46, 6441.

248

H.W. WEBER

Vlcek, B.M., H.K. Viswanathan, M.C. Frischherz, S. Fleshier, K.G. Vandervoort, J. Downey, U. Welp, M.A. Kirk and G.W. Crabtree, 1993, Phys. Rev. B 48, 4067. Wacenovsky, M., 1993, Ph.D. Thesis (Technical University of Vienna). Weber, H.W, 1996, in: Proc. 10th Anniversary HTS Workshop on Physics, Materials and Applications, eds B. Batlogg, C.W Chu, W.K. Chu, D.U. Gubser and K.A. Mtiller (World Scientific, Singapore) pp. 163 166. Weber, H.W., and O. Hittmair, 1979, Supraleitung (Thiemig Verlag, M/lnchen). Weber, H.W, J. Schelten and G. Lippmann, 1973, Phys. Status Solidi B 57, 515. Webel, H.W., G.E Westphal and I. Adaktylos, 1976, Cryogenics 16, 39. Weber, H.W., H. Btck, E. Unfried and L.R. Greenwood, 1986, J. Nucl. Mater. 137, 236. Weinstein, R., Y~ Ren, J. Lin, I.G. Chen, R.P. Sawh, C. Foster and V. Ohot, 1993, in: Proc. Int. Symp. on Superconductivity (Hiroshima), eds T. Fujita and Y. Shiohara (Springer, Berlin) p. 855. Weinstein, R., J. Liu, Y. Ren, I.G. Chert, V Obot, R.R Sawh, C. Foster and A. Crapo, 1994, in: Proc. 1994 Int. Workshop on Superconductivity (Kyoto, Japan), eds S. Tanaka and T. Matsushita (ISTEC, Tokyo) p. 39. Weinstein, R., J. Liu, Y. Ren, R.E Sawh, D. Parks, C. Foster and V Obot, 1996, in: Proc. 10th Anniversary HTS Workshop on Physics, Materials and Applications, eds B. Batlogg, C.W Chu, W.K. Chu, D.U. Gubser and K.A. Mtiller (World Scientific, Singapore) pp. 625-628. Weinstein, R., R.E Sawh, Y. Ren, M. Eisterer and H.W Weber, 1998, Supercond. Sci. Technol. 11, 959. Welp, U., J.A. Fendrich, W.K. Kwok, G.W. Crabtree and B.W. Veal, 1996, Phys. Rev. Lett. 76, 4809. Werner, M., 1997, Ph.D. Thesis (Technical University of Vienna). Werner, M., EM. Sanerzopf, H.W Weber, A. Hoekstra, R. Surdeanu, R.J. Wijngaarden, R. Griessen and K. Winzer, 1996, in: Critical Currents in Superconductors, eds T. Matsushita and K. Yamafuji (World Scientific, Singapore) pp. 203-206. Wiesinger, H.R, EM. Sauerzopf and H.W. Weber, 1992a, Physica C 203, 121. Wiesinger, H.E, EM. Sauerzopf, H.W. Weber, H. Gerstenberg and G.W. Crabtree, 1992b, Europhys. Lett. 20, 541.

Wijngaarden, R.J., H.J.W Spoelder, R. Surdeanu and R. Griessen, 1996, Phys. Rev. B 54, 6742. Wisniewski, A., M. Baran, C. Czurda, H.W Weber, EX. Zhang and L. Zhou, 1994, in: Critical Currents in Superconductors, ed H.W Weber (World Scientific, Singapore) pp. 447-450. Wisniewski, A., C. Czurda, H.W Weber, M. Baran, M. Reissner, W. Steiner, RX. Zhang and L. Zhou, 1996, Physica C 266, 309. Wolf, T., A.C. Bornarel, H. Kiipfer, R. Meier-Hirmer and B. Obst, 1997, Phys. Rev. B 56, 6308. Wtrdenweber, R., G.VS. Sasty, K. Heinemann and H.C. Freyhardt, 1989, J. Appl. Phys. 65, 1648. Workshops on Flux Dynamics, 1994-2000, Workshops held in Palaiseau, Lake Forest, Shoresh, Monte Verita, Hachimantai, Palo Alto, Lunteren. Yacobi, E.R., A. Shanlov, Y. Yeshurun, M. Konczykowski and E Rullier-Albenque, 1992, Physica C 199, 15. Yang, R, and C.M. Lieber, 1996, Science 273, 1836. Yang, R, and C.M. Lieber, 1997, Appl. Phys. Lett. 70, 3158. Yeshurun, Y., E.R. Yacobi, L. Klein, L. Burlachkov, R. Prozorov, N. Bontemps, H. Wtthl and VM. Vinokur, 1994, in: Critical Currents in Superconductors, ed H.W. Weber (World Scientific, Singapore) pp. 237 239. Yeshurun, Y., A.E Malozemoff and A. Shaulov, 1996, Rev. Mod. Phys. 68, 3. Yoo, S.I., N. Sakai, H. Takaichi, T. Higuchi and M. Murakami, 1994, Appl. Phys. Lett. 65, 633. Yoo, S.I., M. Murakami, N. Sakai, T. Ohyama, T. Higuchi, M. Watahiki and M. Takahashi, 1995, J. Electron. Mater. 24, 1923. Yoshida, T., K. Kuroda and H. Saka, 1990, Philos. Mag. A 62, 573. Zandbergen, H.W., R. Gronsky, K. Wang and G. Thomas, 1988, Nature 331, 596. Zhang, EX., L. Zhou, E Ji, W.M. Bian, X.Z. Wn and Z.H. Lai, 1995a, Snpercond. Sci. Technol. 8, 15. Zhang, RX., H.W Weber and L. Zhou, 1995b, Supercond. Sci. Technol. 8, 701. Zhao, Y.J., J.R. Liu, R.L. Meng and W.K. Chu, 1992, Physica C 198, 256. Zhou, L., EX. Zhang, E Ji, K.G. Wang, J.R. Wang and X.Z. Wu, 1990, Supercond. Sci. Technol. 3, 490. Zhn, Y., 1995, in: High Temperature Superconducting Materials Science and Engineering, ed D. Shi (Elsevier, Amsterdam) ch. 5, pp. 199-258. Zhu, Y., H. Zhang, H. Wang and M. Suenaga, 1991, J. Mater. Res. 6, 2507.

FLUX PINNING Zhu, Y., Z.X. Cai, R.C. Budhani, M. Suenaga and D.O. Welch, 1993, Phys. Rev. B 48, 6436. Zhukov, A.A., H. Kiipfer, S.A. Klestov, V.I. Voronkova and V.K. Yanovsky, 1993, J. Alloys & Compounds 195, 479. Zhukov, A.A., H. Kiipfer, S.N. Gordeev, W Jahn,

249

T. Wolf, W.I. Voronkova, A. Erb, G. Mfiller-Vogt, H. W(ihl, H.-J. Bornemama, K. Salama and D.E Lee, 1994, in: Critical Currents in Superconductors, ed. H.W Weber (World Scientific, Singapore) pp. 229-232.

Handbook on the Physics and Chemistryof Rare Earths VoL 31 edited by K.A. Gschneidner,Jr, L. Eyring and M.B. Maple © 2001 Elsevier Science B. g All rights reserved

Chapter 197 MAGNETORESISTANCE AND HALL EFFECT C.C. A L M A S A N 1 a n d M . B . M A P L E 2 1Department of Physics, Kent State University, Kent, OH 44242, USA," 2Department o f Physics and Institute for Pure and Applied Physical Sciences, University o f California, San Diego, La Jolla, CA 92093, USA

Comems List of symbols and acronyms 1. Introduction 2. Upper critical field and irreversibility line of cnprate superconductors 2.1. Upper critical field 2.2. Irreversibility line 2.2.1. Scaling invariance and universality 2.2.2. Origin of the cross-over behavior

251 252

3. Hall-effect studies of cuprate superconductors 3.1. Anomalous sign change in the vicinity of T~ 3.2. Scaling behavior 3.3. Normal-state Hall effect 4. Summary and conclusions Acknowledgments References

252 253 259 259 260

263 263 271 272 276 276 276

List of symbols and acronyms a eL FC H H01 H~2 Hit r

Jo l M ni~ R RH

s T To

T/To T*

intervortex distance Lindemann number field-cooled applied magnetic field lower critical field upper critical field irreversibility line critical current density transport mean free path magnetization Hall carrier number resistance Hall coefficient interlayer distance temperature critical temperature reduced temperature irreversibility temperature

TN

N6el temperature

U

pinning potential

(u2),h

mean-squared thermal vortex fluctuation displacement

V

volume of unit cell

VH or

Hall voltage

Vx~ longitudinal voltage WHHM Werthamel~Helfand Hohenberg Maki ZFC

zero-field-cooled

7

anisotropy parameter

A

magnitude of maximum value of the negative Hall resistivity

OH tc

Hall angle Ginzburg-Landau parameter

),

penetration depth

;t~o

spin-orbit scattering parameter coherence length

251

252 p py. as.y a~,

C.C. ALMASAN and M.B. MAPLE resistivity off-diagonal matrix elements of the resistivity tensor Hall conductivity superconducting contribution to H~ll conductivity

a~y ra ~ug ¢0

normal-state contribution to Hall conductivity cycloton relaxation lifetime transport scattering lifetime magnetic susceptibility magnetic flux quantum

1. Introduction One of the most important and challenging problems in condensed matter physics is the occurrence of superconductivity at high temperatures in the layered copper oxide compounds. Both the superconducting and normal-state properties of these materials are highly unusual and appear to be inextricably linked to one another. A rich and complex temperature vs. carrier concentration phase diagram has been established from experimental investigations of a number of copper oxide systems. Regions in this phase diagram that have been identified up to this point contain antiferromagnetism, superconductivity, intra- and inter-CuO2 plane insulator-metal transitions, Fermi and non-Fermi liquid behavior, a normal phase characterized by a pseudogap, spatially inhomogeneous static and dynamic charge and spin "stripe" phases, and, possibly, a quantum critical point. In the superconducting state, the magnetic field vs. temperature phase diagram has been found to contain a plethora of vortex phases and phenomena, including the melting of vortex lattice and glass phases, activated and quantum flux creep, decoupling of vortices in adjacent CuO2 planes, "peak effects" in the critical current density, dynamic vortex phases, etc. In this chapter, we review magnetoresistance and Hall-effect measurements on high-T~ cuprate superconductors in both the normal and superconducting states that address several issues: for the normal state, the temperature dependence of the Hall coefficient RH and the cotangent of the Hall angle cot OH; for the superconducting state, the temperature dependence of the upper critical field Hc2 and irreversibility line H* as well as the anomalous sign change of the Hall effect near the superconducting critical temperature To. These measurements provide important information regarding the electronic structure and excitations that underly the normal and superconducting properties, fundamental superconducting state parameters, as well as vortex pinning and dynamics for these remarkable materials.

2. Upper critical field and irreversibility line of cuprate superconductors The performance of high-temperature superconductors in magnetic fields is intimately related to the upper critical field Hc2(T), the irreversibility line H i r r ( T ) , and the critical current density Jc(H,T). The Hc2(T) curve defines the region in the H - T plane in which superconductivity occurs and provides information about the nature and origin of superconductivity (i.e., spin and orbital angular momenta of the pairs, pairing mechanism)

MAGNETORESISTANCE AND HALL EFFECT

253

and certain parameters such as the superconducting coherence length ~ and spin-orbit scattering parameter tl~o. The Hirr(T) curve delineates the superconducting region in the H - T plane in which Jc ~ 0 and the magnetization M(H,T) exhibits irreversible behavior. This curve contains information about the phases and dynamics of the vortices in the mixed state between Hol(T), the lower critical field, and H~2(T) (e.g., melting of vortex lattice or glass phases, activated flux motion, coupling between pancake vortices in adjacent conducting CuO2 planes), as well as relevant parameters such as ~, the penetration depth )L, and the pinning potential U, depending on the model used to describe Hirr(T).

2.1. Upper critical field In the early stages of research on high-To oxide superconductors, the Hc2(T) curve was generally inferred from measurements of resistive superconducting transition curves in magnetic fields. As an example, the inset of fig. 1 depicts the normalized electrical resistivity p/p(lOOK) vs. temperature for a polycrystalline sample of YbBazCu307 6 in various applied magnetic fields between 0 and 9 T (Maple et al. 1987). The Ho2(T) data were obtained by defining To in the applied field as the temperature at which p drops to 0.5 of its extrapolated normal-state value. Figure 1 shows two H~2(T) curves that have been fitted to the Hcz(T) data between 0 and 9 T that are derived from the standard Werthamer, Helfand, Hohenberg, and Maki (WHHM) theory (Maki 1966, Werthamer, Helfand and Hohenberg 1966) for conventional type-II superconductors in the limits of complete (~lso=0) and zero (Zso=8) paramagnetic limitation. In 1989, Nakao et al. (1989) estimated that H~f(0) = 40±5 T and H~C(0)= 110+ 10 T from their measurements of hysteretic M - H curves in pulsed magnetic fields above 100 T on singlecrystal specimens of YBa2Cu307 0. Electrical resistivity measurements in megagauss magnetic fields were reported by Goettee et al. (1994) on YBa2Cu307_o epitaxial I

I

F ~ - - - - ~

i

i

1

i

i

i

200 F ; Ba2Cu307_

,0

160 0 4

o12

\

~- 120

" ~ ~ ~

o,I

I ...... H (lesla)

3.5

i

°

O6o.7.oo,o, oo (77K)=30tesla

40

0

20

40 60 80 100 120 TEMPERATURE (K)

Fig. 1. Upper critical field Ho2(T) curves extrapolated in accord with standard WHHM theory from resistively measured Hc2(T) data for YbB~Cu307_ d in magnetic fields up to 9 tesla. Inset: Normalized electrical resistivity p/p(lOOK) vs. T in magnetic fields up to 9T. From Maple et al. (1987).

254

C.C. ALMASAN and M.B. MAPLE 0

o

lOG ZFC 1TZFC

--/~Z~i~

• 2 T FC

• 4 T FC

-0.4

/~/ r/~ .~,/~/ _ ~ y

y/ ,~ 4TZFC * 5TZFC

09 "'1

(.9 v

..~'.,~ .n~oY" ,c. o/"~/ ~ ' /

-0.8 -1.2

-1.6 150

mOT olT E

100

,o

rr

~>5T4'TS2T

oep°7 ~ 2 2 ; ~ ~ ! ¢ ~

~

50 o

og 0 78

80

82

•

~

[]

84

86

88

90

92

T (K)

94

Fig. 2. Upper panel: Temperature dependence of the magnetization of YBa2Cu307_& measured in different magnetic fields oriented perpendicular to the CuO2 planes for zerofield-cooled (ZFC) and field-cooled (FC) measurements. Lower panel: Resistive transition curves for YBa2Cu3OT_a, measured in the same fields as in the upper panel. From Welp et al. (1989).

films. The determination of the Hc2(T) curve with high values of Tc and Hc2(0) from magnetoresistance measurements is, in general, complicated by pronounced broadening of the resistive transition in a magnetic field, which has been attributed to "giant" flux flow (Malozemoff et al. 1988, Tinkham 1988) and/or fluctuation effects in these materials. It has even been conjectured that the resistively determined boundary in the H - T plane may be related more to the irreversibility line than to Hc2(T). Finnemore et al. (1987) were the first to emphasize that Hc2(T) and related thermodynamic parameters for high-Tc oxide superconductors can, in principle, be extracted from measurements of the field and temperature dependence of the reversible magnetization M(/-/, T) using the linear Abrikosov result (Tinkham 1996) 1

4 ~ M ( H , T) -

(2/c 2 _ 1)/3A(Hc2 - H ) ,

(1)

where tc = ~/~ is the Ginzburg-Landau parameter, 3. is the penetration depth, ~ is the coherence length, /~A = 1.16 is the parameter for a hexagonal flux line lattice, and H is the applied magnetic field. Shown in fig. 2 is a plot of the temperature dependence of both the reversible magnetization and resistance measured in different magnetic fields applied perpendicular to the CuO2 planes of YBa2Cu307 6 single crystals (Welp et al. 1989). These reversible magnetization measurements reflect an equilibrium state and, therefore, are not complicated by effects arising from flux motion, as encountered in

MAGNETORESISTANCEAND HALLEFFECT

255

the magnetoresistance measurements. The superconducting critical temperature To(H) is defined as the intercept of the extrapolation of the reversible magnetization in the superconducting state with the normal-state base line. The authors have shown that values of Ho2 and the critical-field slopes for YBa2Cu307_~ single crystals, determined from these M(H, T) data, are about 3 to 4 times larger, depending on the field direction, than those determined resistively. The drawback of the method just described is the following: eq. (1) gives fieldindependent slopes for the M(H, T) curves, i.e.,

OM ) _ 47c ~ f - H

1

dHc2

(2tf2-1)/3a dT

(2)

On the other hand, fig. 2 clearly shows that the slopes (OM/OT)H are field dependent. This implies that the lower temperature region on the reversible M(H, T) curves lies outside the Abrikosov linear regime and, therefore, an analysis of these data in terms of the linear Abrikosov theory is not satisfactory. Most of the high-Tc oxide superconductors have field-dependent slopes (OM/OT)14 in the linear regime and suffer from this drawback. The mean-field theoretical model developed by Hao and Clem allows the anisotropic superconducting state thermodynamic parameters to be determined from measurements of M(H,T) outside of the linear Abrikosov regime (Hao et al. 1991); their model reduces to the Abrikosov linear region only in the vicinity of the transition. However, the determination of the mean-field upper critical field curve Hc2(T) is further complicated by the enhanced effect of thermodynamic fluctuations of the order parameter as a result of the very short coherence length, large anisotropy, and high Tc of the layered oxide superconductors. This presence of thermodynamic fluctuations is readily apparent in magnetoresistance R(H, T) and dc magnetization measurements on these materials. Welp et al. (1991) were the first to determine the mean-field critical temperatures Tc(H) with the magnetic field H applied parallel (H [[ c) to the c-axis for a single-crystal specimen of YBa2Cu3 O7-0 from an analysis of R(H, T) and M(H, T) data using the Ginzburg-Landau fluctuation theory for layered superconductors within the Hartree approximation in the high-magnetic-field limit (Ullah and Dorsey 1991). In contrast to the cuprates with high values of Tc and H~2(0), where only the slope of the upper critical field near Tc, (dHcz/dT)To, could be determined, in the cuprates with low values of Tc and Hc2(0) [e.g., electron-doped Ln2 xCexCuO4 y (Ln=Pr, Nd, Sin), hole-doped T12Ba2CuO6, Yi-xPrxBa2Cu307 o, and BizSr2CuOy compounds], a large part of the H-T phase diagram can be explored for values of T and H that are readily experimentally accessible. This renders the low-Tc cuprates attractive candidates for studies of the T-dependence of Hc2 and of the nature of the H-T phase diagram. A general feature of the low-To cuprates is the striking nearly parallel shift of the resistive superconducting transition curves to lower temperatures with increasing H for both H [[ c and H ± c, with a transition width that remains essentially constant, as in conventional superconductors. Shown in figs. 3a and 3b are the resistive transition curves of an electron-doped Sm1.85Ceo.15CuO4 y single crystal (Tc ~ 11.4 K) in magnetic

256

C.C. ALMASAN and M.B. MAPLE .

,

,

Sm

i

|

,

i

i

,

I

'

'

i

,

|

,

,

0% g ~ ! l P a~d ~ " = 0 tesla

~'3

t.}

0

o o O O o

42

C ',t

.

,

.ssCe0.~ 5CUO.y

2

*'"* nOO,

.*

o o ~ o o . o o

6

*

o

Q-1

oo

ow

o

, '

, *

o

'

HJ.c

o o

o

,g__:i/S/: /i:: $ o

T-v

**° *g

o

o

o

~

o

(a) , '

, |

I I

, "

, '

= '

, '

| I

i '

i '

i •

i '

| I

i '

i '

i '

i '

HIIc o :

g 0

o

~"

Oo : : :

3

o-

o

e _

2

o n

o

o o

° o

o o

1

Oo o

0

, o

~

B

~

~

o- ,~

~

o

~ o

o

,

*

~

.

4,0tesla

.

~.o r..v

~ "

o g

1.5 1.0

0.5

o •

,

o

o

:: = °

o ° =o,

o

:::

" °

,2 o

.

o°

o o

o°

0o

*

o,

e

o o oo

o O2o

O

o

o:

o o

Q.

o

©

.

o

0.35

o

0.2

o~

o -

-

i

i

o.o (b)

-1

i

0

,

•

i

l

5

,

!

i

|

10

T(K)

i

i

i

l

15

i

i

,

,

20

Fig. 3. Electrical resistivity p vs. temperature T for a Sm]85Ce0.15CuO 4 y single crystal in applied magnetic fields up to 10T with (a) H ± c and (b) Hllc. Taken from data reported by Dalichaouch et al. (1990).

fields H J_ c and H II c, respectively (Dalichaouch et al. 1990). The absence of the fieldinduced broadening of the transition width with increasing H is, indeed, quite evident. Consequently, it has been argued that the value of/-/c2 is well-defined in this case (Hidaka and Suzuki 1989, Dalichaouch et al. 1990, Suzuki and Hikita 1990) and, hence, is not affected by flux flow. A striking feature of the resistively determined upper critical field Hc2, however, is its unusual temperature dependence; i.e., the Hc2(T) curve has strong positive curvature over a wide temperature range and appears to diverge as T - ~ 0. This unusual behavior of the resistively determined Hc2(T) has been reported for different cuprate superconductors including the electron-doped materials Ln1.85Ce0.15CuO4-y (Ln = Pr, Nd, Sm) (Dalichaouch et al. 1990, de Andrade et al. 1991, Hidaka and Suzuki 1989), the underdoped compounds R1 xPrxBa2Cu307_ ~ (R=Y, Gd) (Maple et al. 1994, Iwasaki et al. 1990) and YBa2Cu3-xZnxOT_,~ (Walker et al. 1995), and the overdoped compounds T12Ba2CuO6 (Mackenzie et al. 1993) and Bi2Sr2CuO6 (Osofsky et al. 1993). This is in contrast with the Hc2(T) curve of conventional superconductors which has negative curvature and saturates at a constant value as T ~ 0. Interestingly, an Hc2(T) curve with

MAGNETORESISTANCE AND HALL EFFECT I

i

I

I

257

12

I

Srn 1.ssCeo.~sCuO4_y 10

I-LLc

3

g

8

HIIc

6

-1-~ 2

-1Po "" (1)

ID 4

Fig. 4. Upper critical magnetic fields a Sml.ssCe0.15CuO 4 y single crystal determined from p(H, T) measurements in applied magnetic fields H IIc and H / c. Solid lines are guides to the eye. From Dalichaouch et al. (1990).

Hc2(T) of 0

0 2

1.0

,

4

•

6 T(K)

8

•

0.5t ~

,

10

12

•

Sml.asCeo.sCuO4-y 1

00 "'""-.

I ?',,

-r -1.0

-'"""

-2.0 -2.5

' o.2s

o'.s

o;Ts -In(1 - T/l'c)

1 2s

1.5

Fig. 5. In[H~2(T)/H~2(O)]vs. - l n ( l - T/Tc) plot of the data from fig. 3 for Sm~85Ce0~sCuO4_y with H I[c, where H~2(0)= 1.83 T. TN(Sm3+) indicates the N6el temperature below which the Sm3+ ions order antiferromagnetically. The straight lines represent least-squares fits to the data. From Dalichaouch et al. (1990).

positive curvature has also been observed for organic superconductors (Oshima et al. 1988). A n example o f Hc2(T) curves obtained from the p(T,H) measured in two orientations o f the applied magnetic field is shown in fig. 4 for Sml.ssCe0 ~5CuO4 y (Dalichaouch et al. 1990). The data reveal a large anisotropy with Hc2 largest for H J_ c, typical o f layered compounds. Both curves have positive curvature throughout the entire temperature range o f the measurements. The more rapid increase o f Hc2 below 5 K for H [I c was attributed to antiferromagnetic ordering o f the Sm 3+ ions, since the increase in Hc2(T) occurs near the N6el temperature TN~-4.8K and a similar effect occurs in the antiferromagnetic superconductor SmRh4B4 (Dalichaouch et al. 1990)• This rapid increase o f Hc2(T) below 5 K is better illustrated by the l n - l n plot o f fig. 5 (Dalichaouch et al. 1990), in which the

258

C.C. ALMASAN and M.B. MAPLE

101 Srnl.85Ceo.15CuO4.y sin I HIIc

o 0

lO0

-i10-1

10-2

T*/Tc = 0.6

i

i

i

r

~ 1 1 1

~

i

1 - T*/Tc

i

i

i

ii

Fig. 6. Log-log plot of the applied magnetic field H vs. (1-T*/Tc), where T* is the irreversibility temperature and T~ is the superconducting critical temperature for Sm].85Ce0.asCuO4 y. The solid lines represent least-squares fits to the data. From Almasan et al. (1991).

normalized upper critical field, Ho2(T)/H~2(O), is shown as a function of 1 - T/T~ for H ][ c. The plot shows that Hc2 has the following power-law temperature dependence: H c ~ ( r ) = Ho

T "~"~

1 - ~]

,

(3)

with m = 1.6 for 5.7 ~< T ~< 11.4 K (0.5 <~ T/Tc <~1) and m = 3 for T < 5.7 K (T/Tc = 0.5). It was previously suggested that positive curvature of the resistively determined Hcz(T) curve may be a general property of high-Tc cuprate superconductors (Cooper, Loram and Wade 1995, Ovchinnikov and Kresin 1995, 1996, Abrikosov 1997). However, the T-dependence of the resistively determined Hcz(T) curve was interpreted as a measure of the T-dependence of the irreversibility line (Almasan et al. 1992a). Indeed, the irreversibility line Hirr(T M ) extracted from zero-field-cooled and field-cooled dc magnetization M measurements on another Sml.ss Ce0.15CuO4 y single crystal (Tc ~ 20 K) for H 1[c measured in various magnetic fields has also positive curvature and the same power-law behavior, with a change in the value of the exponent from m ~ 3/2 for 0.6 ~< T/Tc ~< 1 to a more rapid temperature dependence for T/Tc <~0.6 (Almasan et al. 1991) (see fig. 6). This result suggests that the H - T line extracted from the R(H,T) measurements on the Sml.ssCe0.15CuO4-y single crystals with H [[ c (fig. 4) is related to the irreversibility line (since it has the same temperature dependence) even for the lower-Tc electron-doped cuprates. In fact, flux-relaxation measurements on a Sml.85Ce0.]sCuO4-y single crystal (Almasan et al. 1991) revealed that the activation energies are comparable to those of the hole-doped YBa2Cu307_~ superconductor (Malozemoff 1989), implying a similar degree of flux motion in both systems. The much weaker broadening observed in the field-dependent resistive transitions in the electrondoped copper oxides would appear to be a result of their lower critical temperatures, not higher pinning energies (Almasan et al. 1991). The main question to be addressed next is the following: if the resistively determined H - T curve even in the low-Tc cuprates is not Hc2(T), than how can one determine Hc2(T)? In an attempt to determine Hc2(T) of the Sm1.85Ce0 ]s CuO4-y compound, Han et al. (1992) analyzed the R(H,T) data in terms of the Ginzburg-Landau fluctuation theory

MAGNETORESISTANCE AND HALL EFFECT

259

100 80 -"

"~

•

" ~

• 'A

H R (kOe)

-

-- WHHMTheory

6O O

\

%

•

"1- 40 "A

20

Hlle

A

•

"A

'ml.85Ce0.15CuO4_y"'A ~,. i

i

r

I

4

,

,

,

I

,

i

i

8

I

12 T (K)

I"l','J',

- ~

.

16

20

Fig. 7. Upper critical field H~2(T ) for Sml ssCe015CuO4_y determined from a fluctuation analysis of R(T,H) data (solid circles). The H~2(T) curve was calculated from the WHHM theory (solid line) and fitted to the Hc2(T ) data, while the irreversibility line Ha(T ) was inferred from R(T,H) measurements (solid triangles). From Hart et al. (1992).

for layered superconductors (Ullah and Dorsey 1991) within the Hartree approximation in the high-magnetic-field limit. As shown by fig. 7, in contrast to the resistively determined R H - T line, which we now refer to as Hirr(T), the Hc2(T) curve extracted with this analysis exhibits negative curvature down to reduced temperatures T/Tc ~ 0.3 and is in excellent agreement with the WHHM theory for conventional type-II superconductors, without paramagnetic limiting effects (i.e., with the spin-orbit scattering parameter )~so= oc). The values of the initial slope (dHc2/dT)r¢=-7.38kOe/K, the in-plane coherence length ~,b(0)=48.4 A, and H~2(0)= 97.6kOe derived from the fluctuation analysis are in reasonable agreement with the values previously obtained by fitting the mean-field Hao-Clem variational model (Hao et al. 1991) to reversible-magnetization data for Sml.85Ce0.tsCuO4 y below T~ and outside of the fluctuation region (Almasan et al. 1992b). 2.2. Irreversibility line 2.2.1. Scaling inuariance and universality A cross-over in both Hirr(T M ) and H~=(T) from an approximate (1 - T/Tc) 3/2 dependence near T/Tc ~ 1 to a more rapid dependence at lower reduced temperatures was first observed in the electron-doped superconductor Sm1.85Ce0.15CuO4-y (Dalichaouch et al. 1990, Almasan et al. 1991, de Andrade et al. 1991), as described above, and subsequently, in both electron-doped (Bud'ko et al. 1990, Sumarlin et al. 1992) and hole-doped compounds, the latter including oxygen-deficient YBa2Cu3Ox (Seidler et al. 1991) and BizSrzCaCu208 + 6 (Kadowaki and Mochiku 1992). The main question was whether the cross-over phenomenon is an intrinsic property of cuprate superconductors (e.g., related to some universal dynamical process involving flux-line mobility or to a dimensional cross-over in the vortex glass/liquid) or a material-dependent effect (e.g., resulting from a sample-dependent distribution of pinning potentials). Almasan et al. (1992a) addressed this issue through measurements of the T-dependence of H~rr(T) in Y1 - xPrxBazCu3 O6.97, a system in which Tc can be systematically reduced from 92 K to 0 K by varying the Pr concentration. On the various resistive transition curves, T~r(H ) was taken to be the

260

C.C. ALMASAN and M.B. MAPLE

temperature at which R(H, T) drops to 50% of its extrapolated normal-state value. The data shown in fig. 8a were obtained by applying a scaling procedure to the Hirr(T R ) data extracted from the R(H,T) curves; for each specimen, H i ar r ( T ) w a s normalized by the cross-over field H*(x), defined to be the value of Hirr(T R ) at T/Tc = 0.6. It is evident that the n o r m a l i z e d HRr(T) data for all x values collapse onto a single curve. The data in the low-field and high-temperature regime follow a power-law dependence with m ~ 3/2. In the high-field regime and for T/To <~0.6, there is a clear departure of the data from the low-field power law; H~r(T)/Ht(x) varies more rapidly than m = 3/2 in this regime, although it is not possible to deduce an explicit power-law dependence. Although the 50% criterion for defining T~rr(H) was arbitrarily chosen in the above analysis, similar results were obtained using the criteria R/Rn = 10% or 90%, where Rn is the extrapolated normal-state resistance. Having established the scaling relation for Y1-xPrxBa2Cu30697, the universality of the scaling conjecture was then tested (Almasan et al. 1992a) by applying it to previously published results on other cuprate systems, notably Sml.85Ce0.15CuO4-y (Almasan et al. 1991, de Andrade et al. 1991; see fig. 8b), oxygen-deficient YBa2Cu3Ox (Seidler et al. 1991; see fig. 8c) and Bi2Sr2CaCu2Os+~ (Kadowaki and Mochiku 1992; see fig. 8d). In each figure, the magnetic field is normalized by a characteristic field H t, which is defined to be the value o f HirM (or HiR or HiXrr, the irreversibility line inferred from magnetic susceptibility X measurements) at T/T~ = 0.6. In all cases, Almasan et al. (1992a) found virtually identical scaling behavior, with a cross-over feature in the universality line occurring at T/T~ ~ 0.6. This result also suggests that the absence of the cross-over in the irreversibility line of fully oxygenated YBa2Cu3Ov 6 is merely a consequence of measurements having not been performed at sufficiently strong magnetic fields for that system. Osofsky et al. (1993) and Mackenzie et al. (1993) reported R(H,T) measurements on Bi2Sr2CuOy thin films and T12Ba2CuO6 single crystals, respectively, and argued that the H - T curves extracted from the magnetoresistance measurements represent Ho2(T). However, the application of the scaling procedure described above to their H - T data reveals that the scaling relation is obeyed here too, as well as the presence of the cross- Hirr(T) R over in at T/Tc ~ 0.6 (see figs. 8d,e).

2.2.2. Origin of the cross-ouer beha~)ior Schilling et al. (1993) observed a cross-over in the irreversibility line measured with H It c on single-crystal specimens of a Bi2Sr2CaCu208 superconductor from a parabolic temperature dependence, H=Ho(1-T/Tc) 2, near To to an exponential dependence, H ~ exp(const/T), at larger H and T/To <~0.5. They argued that the two regimes reflect a cross-over from essentially three-dimensional (3D) vortex fluctuations near Tc to quasitwo-dimensional (2D) vortex fluctuations for H ~>Her ~ 4q~0/$2~2 and low temperatures, where s is the interlayer distance, y =)~¢/&~b is the anisotropy parameter, and q~0 is the magnetic flux quantum.

MAGNETORESISTANCE AND HALL EFFECT 10 2

' x'

• &

101

4,

o • o • o

10o

rr'~; 10.1

' .....

~-o.1 0.2 o3 0.4 o.s 0.53 0.55

i

261

. . . . . .

Ya.x Prx Ba2Cu306.97 ,~

_~,e~ ~ - & I

~

~ m

]

10-2 (a) i

........

i

Sm, ..sCe;j;CuO

101 *~: 10O

,

•

i

i

IIll

'4-'y single cry'st;~l .....

0

H IIc I-~irr(T),grain-alignedsample . ~

•

I'~II~(T), single crystal

1

1

10 4 =

!

10 .2 ........

10 t

YBa2Cu30

i x single

........

,~ roB,

crystal

HIIc

•

.

~

10 0 ~'~ 10.1

~:

10-2

~ ! 0 . 6

101

• A

(c)'

........ i i , I ,i,=, 'M . . . . . . . I ...... iq I-li~r(T),Bi25r2CaCugOs+asinglecPjstal H~('I'). Bi2Sr2CuOyt h i n f i l m ~ H IIc &

10o

~

~

10-1

T/Tc=0.6

10-2

4

(d) , . . . . ., . . ., , . . . . it TI2Ba2CuOs single c r y s t a l

i

. . . . i. .

i

,

,,i,

101 H IIc E

l

'

10 °

10-1 10-2

10-3 10-2

(e) ,

. . . . . . .

i

10-1 1-T/I" c

•

i

,

. . . . .

10 o

Fig. 8. Log log plots of (a) H~., normalized to a characteristic field Ht(x), for polycrystalline Yl__xPrxBa,zCu306.97 (data from Almasan et al. 1992a), (b) H ~ and H~r,., normalized to a characteristic field H l, for a grain-aligned sample and two single crystals of Sml.85Ce0.1sCuO4 y withH IIe (data from Almasan et al. 1991 and de Andrade et al. 1991), (c) H~rr/H t for a single crystal of YBa2Cu3Ox (data from Seidler et al. 1991), (d) H ~ J H 1 and H~./H I for a single crystal of BiaSrzCaCu2Os+ 6 and a Bi2Sr2CuOy thin film, respectively (data from Kadowaki and Mochiku 1992 and Osofsky et al. 1993), and (e) H~. for a T12BazCuO6 single crystal (data from Mackenzie et al. 1993), all vs. 1 T/T c. From Almasan et al. (1992a).

262

C.C. ALMASAN and M.B. MAPLE 105 gm

• 104

. . . . . Bo(To/T. 1 )2

10 2

'

rn

I '

'

'

I '

'

'

I '

'

'

I '

'

'

101 ' o r t e x ~

103

(.9

'

2-D 2-D vortex liquid

--JCLS

E

.,~ 10 o

%

102

~

-~ ~

3-D ~ . ,,vortex liquid

t~

a-D ;'

10 -1

vortex solid~, 101 1 0 -2

(a) ",,

Y1-xPrxBa2Cu306.97

(b)

100

10-3 10

20

30

40

50

60

70

T (K)

80

90

,

0

~

I

0.2

,

,

,

I

0.4

,

,

,

I

0.6

,

,

,

I

,

,

~

0.8

T*/T c

Fig. 9. (a) Vortex-lattice melting temperatures for Bi2Sr2CaCu208 (Schilling et al. 1993) and (b) H I H t ( x ) vs. T I T c for Y~ xPrxBa2CU3OT_~ (Almasan and Maple 1996), both on logarithmic field scales. The solid lines correspond to a fit according to eq. (5). The dashed line in fig. 9a is a quadratic fit, while that in 9b is a power-law fit.

For H > > H c r , the mean-squared thermal vortex fluctuation displacement (U2)th in Josephson-coupled layers is given by

("%,- 8~X~bkBTln(HS272"]

(4)

Assuming that (bt2)th dominates possible quantum fluctuations, Schilling et al. (1993) used the phenomenological Lindemann melting criterion (Lindemaml 1910) to describe the experimentally observed irreversibility line in terms of vortex-lattice melting; i.e., ~b/2) ~ (b/2)th = c2a 2 ,-~

C2~0 H

(5) '

where a is the intervortex distance and CL is the Lindemann number. For H << Her, the Lindemann criterion applied to 3D vortex fluctuations predicts a quadratic T-dependence of the irreversibility line. Almasan and Maple (1996) applied the 3D and 2D vortexfluctuation criteria to the Y1-xPrxBa2Cu306.97 H i R r r ( T ) data. Plotted in fig. 9 on a logarithmic field scale is (a) the vortex-lattice melting temperatures Bim(T) for Bi2Sr2CaCu208 (Schilling et al. 1993) and (b) BRrr/B"(x) vs. T/Tc for the Yl-xPrxBa2Cu30697. system, where Bt(x) is the cross-over field defined above (Almasan and Maple 1996). The solid lines in both figures correspond to a fit according to eq. (5). The dashed line in fig. 9a is a quadratic fit to the data near To, while the dashed line in fig. 9b is a power-law fit of the form B~r/Bt(x ) oc (1 - T/Tc) m with m = 1.45. Recently, Blatter and Ivlev (1993) have shown that the shape of the melting line can be approximated by a power law with an exponent m = 1.45 if quantum fluctuations are included along with the thermal fluctuations

MAGNETORESISTANCE AND HALL EFFECT

263

in the displacement amplitude (eq. 5). Hence, a possible explanation of the discrepancy between the thermal fluctuation result m =2 and the experimental observation m ~ 1.5 for the Y1-xPrxBa2Cu306.97 system as well as for other compounds (e.g., YBa2Cu3Ox and Sml.85Ceo.15CuO4 y) could be that in these systems the quantum fluctuations are important. However, in the Bi- and Tl-based compounds, the thermal fluctuations are enhanced due to the weak interlayer coupling, while the quantum fluctuations remain essentially unchanged; hence, the quantum correction is less important and the data are better described by a power law with exponent m = 2.

3. Hall-effect studies of cuprate superconductors

3.1. Anomalous sign change in the vicinity of Tc The Hall effect in the mixed state of the high-temperature superconductors is one of the most puzzling and controversial transport phenomena exhibited by these materials. In particular, a lot of effort has been devoted to developing an understanding of the unusual sign change of the Hall voltage VH just below the superconducting transition temperature of a wide variety of superconducting materials (Galffy and Zirngiebl 1988, Hagen et al. 1990a, Wang and Ting 1991). Typical Hall voltage Vxy vs. applied magnetic field B data in the neighborhood of Tc for a YBa2Cu307 ~ single crystal, measured with B parallel to the c-axis, are shown in fig. 10 (Almasan et al. 1994a). In general, the Hall voltage Vxy is obtained from the antisymmetric part of the transverse voltage (measured while sweeping the applied magnetic field at constant temperature) under magnetic field reversal. Here, Tc is defined as the lowest temperature at which Vxy is linear in B and extrapolates to zero for B = 0, as is characteristic of the normal state. While V~y is positive and linear in B for T > To, it displays the "negative Hall anomaly" a few degrees below Tc; i.e., with increasing B, Vxy is zero for low values of B, decreases to negative values, passes through a minimum, 15

lr !

10

,

,

~o.1

.

.

.

~o[ --./,.. / 1 o ~ > t "~1[ ~ f ~ "@" 1 "1o ':~ ' ~ ' "6 ' ~ ° ~ i ~ " " /

.--. > '~ >

.

T=92K

H (T)

87.5t' 86K 83K

L

0 -5

~

:

~,.,t.- ~ I

0

I

2

I

....

I

4 H (T)

/

I

81K 76K

~

I

6

I

I

8

Fig. 10. Hall voltage Vxyvs. applied magnetic field H for an YBa2Cu3OT_ 6 single crystal with H II c at several temperatures in the neighborhood of T~. Inset: Hall voltage V~, and longitudinal voltage V~x vs. H at 81K. From Almasan et al. (1994a).

264

C.C. ALMASAN and M.B. MAPLE

0.8. "0i . . . . . . . . . . . . . . . . .

fot 0o.co

0 6

E

~

1 xCOx) a

o~.ee

o

.4

~" " ' ~

[

/

""

x0

<, Y B a 2 ( C u

ifl

o~

x=O'08"

• o

x=013 y=675

I "%,, °~ ,,,~ •

c~c><~,--~

I

08

.

.

.

.

I

09 T/T C

.

y=673_ (x>O 24

ao4oso6oTosogo %

. . . .

07

~ ~

~'

~,~

.

,

.

!

1

. . . .

11

Fig. 11. Magnitude of the maximum, A, of p,y(H) for Y1 xPr~Ba2Cu307 ~, YBa2Cu3Ox, and YBa2(Cu 1 xCoa)307_,3 single crystals, taken at different temperatures, vs. reduced temperature T/T~. The data exhibit a maximum Am,× which is plotted in the inset as a function of T~. Also shown in the inset are two data points for YBa2(Cu I ~C%)307 ~. From Almasan et al. (1995).

crosses zero, and then becomes positive for values of B lower than Bc2(T). The region of negative Vxy shifts to higher fields at lower temperatures. Shown in the inset to fig. 10 are V~, and the longitudinal voltage Vxx vs. B, measured at 81 K. The onset of the negative Hall signal occurs at a slightly higher field than the onset of Vx~, while the negative Vxy extends over a region where Vxx increases rapidly with increasing field. Similar Vxy vs. B profiles were reported in most of the other optimally doped hole-like charge carrier cuprates: e.g., YBa2Cu307 a (Galffy and Zirngiebl 1988, Iye, Nakamura and Tamegai 1989, Hagen et al. 1990a, Chien et al. 1991a, Luo et al. 1992, Rice et al. 1992, Kunchur et al. 1994), Bi2Sr2CaCu2Ox (Iye, Nakamura and Tamegai 1989, Zavaritsky, Samoilov and Yurgens 1991), and T12Ba2CaCu208 (Hagen et al. 1991a, Samoilov, Ivanov and Johansson 1994), and electron-like cuprates Nd185Ce0.15CuO4 y (aside from the overall sign which is reversed) (Hagen et al. 1993). The existence of a similar Hall anomaly in both n- and p-type materials provides a useful test for theoretical models of the V~y(B) dependence. Additionally, similar V~y vs. B profiles were reported in YBa2Cu3OT/PrBa2Cu307 multilayers (Qiu et al. 1995) and (YBa2Cu307_a)n/(PrBa2Cu307 a),, superlattices (Kebin, Yuheng and Adrian 1996). In addition, similar Vxy vs. B profiles have been observed in some underdoped cuprates; e.g., Yl-xPrxBa2Cu307-6 single crystals with 0 ~<x < 0.13 (Jia et al. 1993) and YBa2Cu307 a with 1 or 2% of the Cu atoms replaced by Co (Neiman et al. 1995), while only a positive Hall signal has been observed in other underdoped cuprates; e.g., YBa2Cu307 b with 3 or 4% of the Cu atoms replaced by Co (Ginsberg and Dudey 1993, Neiman et al. 1995). A systematic study of the evolution of the magnitude of the maximum value of the negative Hall resistivity Pxy with To has been carried out by Almasan et al. (1995) on two cuprate systems, Ya-xPrxBa2Cu307 d and YBa2Cu3Oy, over a wide range of doping in order to determine if there is a correlation between the negative Hall anomaly and the doping level. Figure 11 is a plot of the magnitude of the maximum value A of the negative Hall resistivity pxy, extracted from Pxy vs. B curves (see the inset to fig. 10 for the definition of A), vs. the reduced temperature T/Tc for

MAGNETORESISTANCEAND HALLEFFECT

265

underdoped YI xPrxBa2Cu307 6 and YBa2Cu3Oy single crystals (Almasan et al. 1995). The Pr content was estimated by comparing the transition temperatures of the single crystals to those of high-quality polycrystalline samples (Maple et al. 1992, Paulius et al. 1994). The legend lists the samples in decreasing order of Tc. For x = 0 , a similar result was reported by Luo et al. (1992) for epitaxial YBa2Cu307 films. The two interesting features of all these curves are: (1) for both systems, A is nonzero over the same reduced temperature range (0.87 < T/To <<.1) with the onset of pinning occurring at T/Tc ~ 0.87 for all the single crystals of Y1 -xPrxBa2Cu3OT_ ~5with 0 ~<x < 0.13 and YBa2Cu30y with 6.73 < y ~<6.84, and (2) Z~max scales with Tc and decreases monotonically, as shown in the inset. A linear fit of the Amax vs. Tc data with Amax ~ 0 extrapolates to Am~x= 0 at Tc = 75.2 K. This suggests that pxy would not display the negative anomaly for samples with Tc <75 K. With few exceptions (Manson et al. 1996), the data available in the literature seem to agree with this finding (Ginsberg and Dudey 1993, Neiman et al. 1995). Certainly, the criterion used to define Tc is an important factor in this analysis. With regard to feature (1), it is intriguing to note that the reduced temperature T/Tc ~ 0.87, below which z~ vanishes, is very close to the value of T/Tc ~ 0.85 reported by Budhani, Holstein and Suenaga (1994) from resistivity measurements on T12Ba2CaCu208 heavy-ion-irradiated epitaxial films, below which pinning by columnar defects inhibits flux motion and, hence, dissipation. Although the depinning temperature varies from material to material, these results suggest that the onset of pinning occurs at the same reduced temperature, at least in the three systems mentioned, regardless of the nature of the pinning site (point or columnar defects). A survey of the literature (Hagen et al. 1993) has indicated that, in general, the anomalous Hall behavior is present among moderately clean superconductors (l/~ ~ 0.5-5, where l is the transport mean free path and ~ is the coherence length), but does not occur in either the very clean (l/~ >> 1) or dirty (1/~ << 1) limits. Colino et al. (1994) have reached the same conclusion by studying epitaxial RBa2Cu307 (R = Eu, Ho) thin films with the c- or a-axis perpendicular to the film plane. Hence, the parameter 1/~ seems to largely determine whether or not the Hall anomaly is observed. Recently, it has been shown theoretically that, indeed, in the very clean limit, the Hall conductivity is similar to that for the normal state in the limit of high fields, while, in the moderately clean case, the sign of the Hall effect in the mixed state may differ from that in the normal state depending on the detailed structure of the Fermi surface (Kopnin and Lopatin 1995); therefore, these authors have concluded that the sign reversal is very sensitive to the details of the Fermi surface structure, the doping level, and the impurity content. Several interpretations have been offered to explain the sign reversal of the Hall voltage of the mixed state. Freimuth, Hohn and Galffy (1991) attributed the sign reversal to the large thermoelectric effects in the mixed state. Harris, Ong and Yan (1993) reported the observation of opposite Hall signs for vortices perpendicular and parallel to the CuO2 layers of YBa2Cu307_o single crystals. The competition between the two opposing Magnus forces acting on interlayer segments and pancakes which form the vortex line, respectively, led them to an explanation of the negative Hall anomaly in terms

266

c.c. ALMASANand M.B. MAPLE

of "pancake" vortices which lie in the CuO2 planes and are linked by interlayer segments which reside in between and parallel to the CuO2 planes. Another point of view, which has lately attracted more supporters, is that this behavior is an intrinsic property (Hagen et al. 1990a,b, 1991a,b, 1993, Almasan et al. 1995) and should have an explanation in terms of the general theory of flux motion in type-II superconductors. This is corroborated by the fact that the sign reversal was observed earlier for conventional V and Nb superconductors (Niessen, Staas and Weijsenfeld 1967, Noto, Shinzawa and Muto 1976). The observed field independence of the sign-reversing Hall angle (in the isolated vortex limit B --~ 0) in BizSr2CaCu2Ox and TlzBa2CaCu208 (Hagen et al. 199la, Zavaritsky, Samoilov and Yurgens 1991) which is expected in the vortex dynamics picture but not within other models, is strong evidence for such an intrinsic origin of the Hall anomaly. The close similarity of the profiles of fig. 11 and the monotonic evolution of Amax with Tc for the underdoped Y1 xPrxBa2Cu307 6 and YBa2Cu3Oy systems indicate also that the negative anomaly has an intrinsic origin (Almasan et al. 1995). Additionally, measurements of pxy(H) in the mixed state of Ndl.85Ce0.15CuO4-y (Hagen et al. 1993) provide support for the flux-dynamics interpretation, as well as evidence against thermoelectric models (Freimuth, Hohn and Galffy 1991) which attempt to explain the Hall anomaly in the mixed state of the cuprates. The early theories of flux motion (Bardeen and Stephen 1965, Nozi~res and Vinen 1966) considered the Hall effect in the mixed state as being produced by normal electrons in the vortex cores and thus predicted an effect of the same sign as in the normal state. The first microscopic calculations of the flux-flow Hall effect in clean superconductors used a simple parabolic spectrum of electrons with a single effective mass, and could not provide a comprehensive picture of the sign of the Hall effect (Kopnin and Kravtsov 1976a,b). Further progress has been made only recently on the basis of time-dependent GinzburgLandau equations derived for a gapless superconductor in the dirty limit (Dorsey 1992, Kopnin, Ivlev and Kalatsky 1993). The main result is that the sign reversal can be intrinsic and depends on the details of the structure of the normal-state electronic spectrum. This conclusion was further confirmed by microscopic calculations for dirty superconductors (Larkin and Ovchinnikov 1995). This picture has found experimental confirmation (Jones, Christen and Sales 1994, Kunchur et al. 1994, Samoilov, Ivanov and Johansson 1994). Very recently, within a phenomenologinal analysis, Feigel'man et al. (1994) and Khomskii and Freimuth (1995) considered an additional contribution to the Hall effect due to the positive difference between the electron density at the center of the vortex core and that far outside the vortex. This contribution has a sign that is opposite to that of the conventional one and can cause a sign change. Hence, these researchers interpret the sign change in terms of broken particle-hole symmetry and obtain good agreement with the experimental signatures (Hagen et al. 1993) of this effect. Van Otterlo et al. (1995) have recently presented a microscopic derivation of the equation of motion for a vortex in a superconductor. They have obtained a coherent view of the vortex dynamics in which both hydrodynamics and the vortex core contribute to the forces acting on a vortex. The competition between these two provides an interpretation of the observed sign change in the Hall angle in superconductors with mean free path l of the order of the coherence

MAGNETORES1STANCEAND HALLEFFECT

267

length ~ in terms of broken particle-hole symmetry in the electronic band structure, yielding a unifying description of the physics involved. Other theoretical explanations for this anomaly have also been put forward. Ferrell (1992) proposed that the negative Hall sign arises from the "upstream" force exerted by backflow of quasiparticles on the core, two-band models were proposed by Hirsch and Marsiglio (1991) while Wang and Ting (1991) and Wang, Jinming and Ting (1994) have proposed that the sign reversal results from the interaction of moving vortices with imperfections of the crystal structure. However, this latter proposal is refuted by several experimental findings (Budhani, Liou and Cai 1993, Almasan et al. 1995). Specifically, the suppression of Amax with decreasing T~ [see inset to fig. 11] suggests that the negative Hall anomaly is not due to flux pinning since, in Y1 xPrxBa2Cu3Ov-a with low x values, the pinning potential increases with increasing x (decreasing Tc) (Paulius, Almasan and Maple 1993). Within the context of flux-pinning models, this would produce an increase in the magnitude of the negative Hall voltage. Also, the reported decrease in the magnitude of the Hall effect sign anomaly of T12BazCazCu3Ol0 thin films with increasing defect concentration through heavy-ion irradiation (Budhani, Liou and Cai 1993) is inconsistent with the attempt to explain the negative Hall anomaly in terms of the presence of pinning forces in the sample. Although Hall-effect results for superconductors have traditionally been expressed in terms of the Hall coefficient Rm the Hall angle OH, and the off-diagonal matrix elements of the resistivity tensor Pox, it is of particular interest to discuss the Hall effect in the mixed state in terms of Oxy, since Vinokur et al. (1993) pointed out that this quantity should be independent of the disorder in the sample. Ullah and Dorsey (1991), Dorsey (1992), Kopnin, Ivlev and Kalatsky (1993) and Geshkenbein et al. (1994) have shown that within the time-dependent Ginzburg-Landau formalism, the Hall conductivity in the superconducting state can be expressed as the sum of superconducting, OSy, and normal, one, contributions: CI ~ CzB,

(6)

where

Oxy

Pox PxxPoy - PxyPyx

V,yd/I PxxPov- PxoPyx

(7)

with d the thickness of the sample, I the applied current, and PO (i,j=x,y) the matrix elements of the resistivity tensor. In eq. (6), 6sy arises from the motion of the magnetic vortices, while o~v is associated with the motion of the quasiparticles in the vortex cores. If the Hall currents produced by the motion of the vortices and the quasiparticles (C1 and C2, respectively) have opposite signs, then the Hall effect can change sign as B is varied, as observed experimentally. The sign of o~ is the same as the sign in the normal state. However, the factors that fix the sign of 6~:v are not yet understood. Although a few microscopic calculations have addressed this issue (Vinokur et al. 1993,

268

C.C. A L M A S A N and M.B. MAPLE 20.0

.

.

.

.

i

.

.

.

.

i

.

.

.

.

i

.

.

.

.

,

.

.

.

.

,

.

.

.

.

10.0 0.0 -10.0

Q

E -20.0 = -30.0 I.~o . . . .

•

o°

o

o

T=79

*

T=81.02K

07K

(a) i

. . . .

.

.

.

.

.

i

.

. . . .

.

.

.

.

I

.

. . . .

.

.

.

.

i

.

.

.

.

.

.

,

i

.

.

.

. . . .

.

.

o

m

7.0 6.0 5.0

o

4.0

--

T=39.01K

~,---~ T = 4 2 . 0 4 K

(b)

3.0 20

25

30

35 40 B2 (T2)

45

50

Fig. 12. o~,Bvs. B 2 data for (a) Yo.92Pr0.0aBa2Cu307 6 and (b) YosaPro42Ba2Cu3Ov_,~ single crystals measured at different temperatures. The solid lines are fits to the data with eq. (6). From Almasan et al. (1995).

Feigel'man et al. 1994, Kopnin 1996), there is considerable uncertainty about why, in certain cuprates, o~y is negative while oxnyis positive. Several groups have used eq. (6) to analyze the Hall conductivity data of different high-temperature superconductors in order to ascertain whether eq. (6) provides a satisfactory description of the Oxy(B) data, and, if it does, to determine how the two coefficients C~ and C2 depend on temperature and doping. Figures 12a and 12b are typical plots of axyB vs. B 2 for specimens which do and do not, respectively, display the negative Hall anomaly. These graphs depict data taken on single crystals of Yo.92Pro.08Ba2Cu307-6 (fig. 12a) and Yo58Pro.42Ba2Cu307-6 (fig. 12b) at various temperatures T < Tc and in applied magnetic fields B m ~< B ~<5 T, where Bm is the field below which the Hall resistivity starts bending over, presumably due to flux pinning (Almasan et al. 1995). Also shown in the figures are straight-line fits of eq. (6) to the data which give values for the coefficients C1 and C2 which are independent of B but dependent on T. Whereas C1 < 0 and C2 > 0 in fig. 12a, Cl and C2 are both positive in fig. 12b, consistent with the presence of the negative Hall anomaly in Pxy of the sample (fig. 12a) or its absence (fig. 12b). The results for the Yo.92Pro08Ba2Cu307-6 single crystal are similar to those obtained on other single crystals which display the Hall anomaly in the mixed state; i.e., YBazCu307 6 (Clinton et al. 1995, Ginsberg and Manson 1995, Harris et al. 1995) and NdtasCe0.15CuO4 (Lobb et al. 1994, Clinton et al. 1995). However, the Hall conductivity data for YBa2Cu3Oy ( y = 6.84, 6.75) single crystals (Almasan et al. 1995) and La2 xSr~CuO4 thin films (Matsuda et al. 1995), which also display the negative Hall anomaly, are described well by the time-dependent Ginzburg-Landau formalism with

MAGNETORESISTANCE AND HALL EFFECT

269

1.5

Yl_xPrxBa2Cu307_g YBa2Cu3Oy

1.0

/ /~v

/

'or 0.50 'T,

g

I,0

o~ -20 -40 -60

~-I~x= 0 \ x=O,08 \ ,t

-80

~

x = 0.13

y = 6.75 •-x=0.24

-~v-~ x = 0.29 -100

•

x=o.~2,

\.\ ~ \ \

\

\

\

\

\

0,05

\

~. \ "~ \ \ \

\

0.1 1-T/T °

\ 0.1~

Fig. 13. Coefficient Ci vs. 1 T/Tc, for six Yl-~Pr,-Ba2Cu307 a and two YBa2Cu3Oy single crystals with different values of x and y. The samples with x = 0.13 and y = 6.84 and 6.75 display the negative sign anomaly while those with x > 0.13 do not. The solid lines represent power-law behavior indicated in the figure. From Almasan et al. (1995).

an additional constant term C3 included in eq. (6). It seems that the C3 term becomes important with increasing anisotropy. On the other hand, the Hall conductivity data for T12Ba2CaCu2Q and Bi2Sr2CaCu2Os+a, which display the negative Hall anomaly, are described well by eq. (6) with C2 = 0 and an additional constant term C3 (Samoilov, Ivanov and Johansson 1994). The dxy(B) data for YBa2Cu3Oy ( y = 6.73, 6.65) show an even more complicated field dependence which seems to be sample dependent (Almasan et al. 1995). The results of fig. 12b are similar to those obtained on specimens which do not display the negative anomaly; i.e., single crystals of YBa2Cu307 ,~ with 3% and 4% of the Cu atoms replaced by Co (Ginsberg and Dudey 1993, Neiman et al. 1995) and single crystals of Lal.s3Sr0.wCuO4 (Harris et al. 1995). There have been several studies which report the temperature dependence of the two coefficients Cl and C2 for different cuprate superconductors. As discussed below in more detail, these results seem to indicate that C1 has the same behavior in most of the cuprates studied while the behavior of C2 is sample and/or system dependent; i.e., the power-law temperature dependence of C1 is the same for all the specimens studied with an exponent a decreasing as the samples are more underdoped, while C2 displays temperature dependences which differ from system to system. Shown in fig. 13 is a plot of CI vs. 1-T/Tc for six YI xPrxBa2Cu3OT_~ and two YBa2Cu3Oy single crystals with different values o f x and y (Almasan et al. 1995). The legend lists the samples in order of decreasing To. In the superconducting state, the coefficient C1 follows a systematic trend with To and is negative (positive) for samples which display (do not display) the negative pxy anomaly. Over the same reduced temperature,

270

C.C. ALMASAN and M.B. MAPLE 1.4

,

i

i

i

1.2 -Yl_xPrxBa2Cu307. 8 ~x •

1.0 0.8

x=O

zx

x=0.08

•

x=0.13

• v

x=0.24 x=0.29 x=0.42

•

0.6

~' ~x • q~x ~ •••

,,

E 0.4

o

0.2

v,~•"A~v-'v v v

--t "7

I

I-~:~ 0.6 o

••v

I

YBa2Cu3Oy

• • • I

~

©,U

O

Oo

0.2

~/

' I

°]

40l ©

I

'E o y = 6.84 o ° 8 ° ho y = 8 . 7 5 ~

0.4 0.3

• I

~"12'01

y=6.84 <> Y=6"75 °~ o

0.5

••

I

~ ~

-0.

o°/°1 -0.1

TO 0.1 1- fr c

0,2

o° °

0.1

o .o o

0.0

,,,

0.8

I

I

0.9

1

I

I

I

1.1

1.2

1.3

Tc/T

1.4

Fig. 14. Coefficient C2 vs. TJT, for six Y, _yPr~Ba2Cu307. ~ and two YBa2Cu3Oysingle crystals with different values of x and y. Inset: C~ vs. 1- T/Tc, From Almasanet al. (1995).

the magnitude of C1 is much larger for the samples which exhibit the negative anomaly (note the change in scale for C1 < 0 and C] > 0 in fig. 13) with an equally good fitting of the data to eq. (6) for low and high values o f x (see figs. 12a,b). The coefficient C1 was found to have the form C1 =A(1 - T/Tc) a with the exponent a decreasing with increasing doping level; a ranges from 2.3-t-0.07 for y = 6.84 to 1.44-0.07 for y = 6.75, in the range of x and y where CL < 0, and from 2.8±0.05 for x = 0.24 to 1.44-0.06 for x =0.42, in the range o f x where Cl > 0. As expected, C1 vanishes within experimental error for T ~> Tc. Ginsberg and his collaborators have found that for YBa2(Cul_xCox)307_~ (x=0, 0.01, 0.02, and 0.03) single crystals (Ginsberg and Manson 1995, Neiman et al. 1995), detwinned Yl xPrxBa2Cu3OT_~ single crystals (Manson et al. 1996), and YBa2(Cul ~Nix)3Ov-,5 single crystals (Jin-Tae, Giapintzakis and Ginsberg 1996), the coefficient Ct follows the same power law temperature dependence as presented above for twinned single crystals of Yl-~PrxBaaCu307 ,5. The exponent a has a value of 2 for all the samples of YBa2(Cul-xCo~)307-6 studied, within experimental uncertainty, with a hint that it may be increasing with increasing cobalt concentration, while the value of a for Pr and Ni doping decreases with increasing doping level. Figure 14 contains a plot of C2 vs. Tc/T for six YI-~Pr~Ba2Cu307-~5 (upper panel) and two YBa2Cu3Oy (lower panel) single crystals; the inset of the lower panel is a plot of C3 vs. 1 - T / T c for the two YBa2Cu3Oy single crystals. For Y] xPrxBa2Cu3OT_d, C2 in the superconducting state is positive and approximately proportional to Tc/T for all r values with a slope which decreases with increasing x, while, in the normal

MAGNETORESISTANCE AND HALL EFFECT

271

state, C2 is positive, small, and has a weaker temperature dependence relative to the superconducting state (see upper panel). For the two YBa2Cu3Oy single crystals, C2 increases with decreasing T, displays a maximum for T = T~, and then decreases with decreasing T (see lower panel). The coefficient C3 is positive in the superconducting state and decreases with increasing T to zero within experimental error at T = Tc (see inset). On the other hand, Ginsberg and coworkers have found that C2 = a + b(1 - T/T~) for YBa2(CUl xCox)307 ~ single crystals (Ginsberg and Manson 1995, Neiman et al. 1995), while C2 = a + b(1 - T/To)2 for detwiuned Y1 - xPrxBa2 Cu3 07 - ~ single crystals (Manson et al. 1996) and YBa2(Cul - x N i x ) 3 0 7 c5 single crystals (Jin-Tae, Giapintzakis and Ginsberg 1996). 3.2. Scaling behavior An interesting scaling behavior between the Hall resistivity Pxy and the longitudinal resistivity Pxx, Pxy c~P~xx, has been observed with the scaling exponent / 3 ~ 2 in Bi2Sr2CaCu208 single crystals (Samoilov 1993), T12Ba2CaCu2Oa (Samoilov, Ivanov and Johansson 1994) and T12Ba2Ca2Cu3Ol0 (Budhani, Liou and Cai 1993) films, and (YBa2Cu307 6)n/(PrBa2Cu307 6)m superlattices (Kebin, Yuheng and Adrian 1996) and YBa2Cu307_JPrBa2Cu307_(~ multilayers (Qiu et al. 1995), and /3= 1.5-2.0 in YBa2Cu307 6 films (Luo et al. 1992, Kang et al. 1996). The scaling behavior p~y ocp~x has been observed in the regime of positive Hall effect for some cuprates and negative Hall resistivity for other cuprates. Kebin, Yuheng and Adrian (1996) have found that the sign anomaly in (YBa2Cu307 6)n/(PrBa2Cu3Ov ~5)msuperlattices diminishes and even disappears with decreasing YBa2Cu307 ~ layer thickness, i.e., increasing anisotropy, while the scaling behavior holds for all the samples, regardless of the anisotropy value, with some variation in the exponent (/3=1.6-2.0). On the other hand, Wang, Yang and Horng (1997) have found that the sign reversal of the Hall coefficient of (YBa2Cu307_~)J(PrBa2Cu307_cS)m superlattices diminishes with decreasing YBazCu3Oy_d layer thickness or increasing PrBazCu307 layer thickness, while the exponent/3 is magnetic-field dependent and is larger for a (YBa2Cu3Ov_,~)J(PrBa2Cu307_ ~)m superlattice with a thinner YBa2Cu307 ~ layer. All these findings seem to rule out a possible link between scaling and sign change of the Hall resistivity. An example of the scaling between the Hall and longitudinal resistivities is shown in fig. 15, a log-log plot of Hall resistivity IPxyl vs. longitudinal resistivity px~ of an epitaxially grown YBa2Cu307 film (Luo et al. 1992). The data shown were obtained for temperature sweeps at two magnetic field values of 1.4 and 3.7 T. The data display the power-law relationship in the negative Hall regime, with the exponent/3 = 1.74-0.2 for both fields. The cusp is an artifact of the absolute value ]Pxy], corresponding to where pxy(T) passes through zero. The power-law relationship breaks off abruptly at high temperatures, perhaps signaling the transition to flux flow and/or the normal state. Similar results were obtained for all the other cuprate superconductors for which the scaling relationship was observed.

272

C.C. ALMASAN and M.B. MAPLE

-1 -->. ×

B=I.4T

B=3.7T

1

-2

-%-3 .00

.

-4 2

-5

j

-~ " .

• ' • Pxy < 0

A~

o,z, Pxy>O

A

-6

-0.50

0'.0

0;0

1'0

log Pxx

115

2.0

Fig. 15. Log log plot of [px~[ us. pxx data for epitaxial YBa2Cu307 films obtained for temperature sweeps at two magnetic field B values. Units for p are #tf2cm. The straight lines are fits of the data with IP~[ ecp~, where /3= 1.7±0.2 for both fields. From Luo et al. (1992).

Dorsey and Fisher (1992) showed that the scaling behavior p~y =Apex is a general feature near the vortex-glass transition, but they were not able to obtain an exact value of the scaling exponent. A model proposed by Vinnkur et al. (1993) suggests that the scaling exponent should be 2 in the thermally assisted flux flow region regardless of the pinning strength. Their analysis suggests that the scaling behavior is of a different origin than the sign change in the Hall resistivity. This is consistent with the experimental data of Luo et al. (1992), Samoilov (1993) and Kebin, Yuheng and Adrian (1996) discussed above. The results of Vinokur and coworkers are in fair agreement with scaling exponents of both weakly pinned systems; i.e., Bi2Sr2CaCu2Q crystals (/3=2.0+0.1) (Samoilov 1993), and rather strongly pinned systems; i.e., heavy-ion-irradiated T12Ba2Ca2Cu3Ol0 films (/3= 1.85-t-0.1) (Budhani, Liou and Cai 1993), but not with the recent results on heavy-ion-irradiated YBa2Cu3OT_`5 (/3 = 1.54-0.1) (Kang et al. 1996). On the other hand, Wang, Jinming and Ting (1994) modified their earlier work (Wang and Ting 1991) to develop a unified theory of the Hall effect, including both the pinning effect and the thermal fluctuations. Wang, Jinming and Ting (1994) explain the scaling behavior and the anomalous sign reversal of the Hall effect by taking into account the backflow current due to pinning. In their model, the scaling exponent changes from 2 to 1.5 as the pinning strength increases. In agreement with this theory, Kang et al. (1996) have found that the scaling exponent of YBa2Cu307 ,5 single crystals changes from 2 to 1.5 as the pinning strength increases through heavy-ion irradiation, while Kang et al. (1997) have found that for HgBa2CaCu206+`5, fi increases from 1.5-t-0.1 to 1.9-t-0.1 as the magnetic field increases from 1 to 5.5 T. The authors suggest that these latter results indicate that the pinning strength decreases with increasing magnetic field. 3.3. Normal-state Hall effect Among the normal-state transport properties of the high-temperature cuprate superconductors, the Hall effect remains one of the most difficult to explain. In the majority of the cuprates, the Hall coefficient RH falls monotonically with increasing temperature. The

MAGNETORESISTANCEAND HALL EFFECT 1.4

. . . .

• Y.

i

1.2 - ,-x

. . . .

Pr Ba

x

I

2

. . . .

Cu

i

. . . .

O_ ~

3 ,-~

I

. . . .

I

.

.

.

.

..~x=O.08

7jo

x=O

1.0

o.8

f

o//....< j

~-0.6

.,, x=o.a9

xoo.,

x=0.42

0.4 0.2 0.0 50

100

150 T(K)

273

200

250

300

Fig. 16. Temperature dependence of the Hall carrier number n H-V/eR H for Y1 ~PrxBa2Cu307-~ single crystals with different x values. The fact that the data for two pairs of x values (x=0, x=0.08, and x=0.42, x=0.51) are out of sequence in the V/eR H vs. T plot, but are not in the cot(OH) VS. T2 plot (see fig. 17), could be a result of the error in measuring the thickness of these crystals. The x=0.9 value represents the nominal composition, while the x values of the superconducting samples were estimated by comparing the T~'s of the crystals with those of high-quality polyerystalline samples. From Maple et al. (1994).

pronounced temperature dependence o f RH is in sharp contrast with the behavior o f RH in conventional metals with complicated Fermi surfaces such as Cu, Ag, Ca, Mg. In these elements, RH becomes insensitive to temperature when the temperature exceeds -0.3OD, where Oi) is the Debye temperature. Yet another puzzling feature o f the Hall effect is the sensitivity o f the anomalous temperature dependence to in-plane disorder. Replacement of Cu by magnetic ions such as Co, Fe, or Ni or by nonmagnetic Zn reduces the slope o f the Hall carrier number na = V/eRH ( V is the unit-cell volume) vs. T as Tc is suppressed (Clayhold et al. 1989). A similar behavior has also been observed in underdoped Y1 _xPrxBa2Cu307 6 (Wu et al. 1992, Almasan et al. 1994a,b) and in Nidoped La2 xSrxCuO4 (Clayhold et al. 1989). An example o f the temperature dependence of nu and its evolution with doping is shown in fig. 16 in which n u ( T ) data are displayed for the Y1 xPrxBa2Cu307_ 6 system with different x values (Maple et al. 1994). Notice that nil(T) is positive and, for x ~< 0.13, shows a remarkably linear temperature dependence in the normal state in the range 100 K ~< T <~ 250 K and actually extrapolates to zero at T = 0 K. Similar results were obtained on other optimally doped YBa2Cu;O7 6 specimens (Clayhold et al. 1989, Iye, Nakamura and Tamegai 1989, Rice et al. 1991, Wu et al. 1992). The upturn just above Tc was interpreted as a signature o f superconducting fluctuations (Rice et al. 1991). With increasing x, the temperature range o f the linear dependence o f n u ( T ) is reduced and the overall magnitude o f nil(T) decreases monotonically (Jia et al. 1992, Wu et al. 1992, Maple et al. 1994). Also, the temperature dependence o f nH appears to become less pronounced with increasing x. A similar behavior o f nil(T) with doping was observed in other underdoped cuprates; e.g., Ni-doped ceramics (Clayhold et al. 1989) and YBa2Cu3 xZnxO7 6 single crystals (Chien, Wang and Ong 1991b). Several interpretations were given o f the observed linear temperature dependence o f the in-plane resistivity and Hall carrier number. Although a strong temperature dependence is possible in a two-band model provided that the scattering rates are different in these two bands, a linear T-dependence requires precise cancellation between electron and

274

C.C. ALMASANand M.B. MAPLE

hole contributions. This is highly unlikely to happen in all cuprate superconductors. Magnetic skew scattering is another mechanism that can cause a strongly temperaturedependent Hall effect (Fiory and Grader 1988), as reported in some heavy-fermion systems (Fert and Levy 1987) and ferromagnetic alloys (Rhyne 1968 and Bergmann 1979). Localized magnetic moments in these systems scatter conduction electrons asymmetrically, causing an anomalous contribution to the Hall effect. However, the fact that the low-temperature Hall resistivity does not saturate argues against such an explanation. In the Luttinger-liquid model of Anderson (1990) the intrinsic electronic degrees of freedom (i.e., charge and spin) are decoupled in the CuO2 planes, with each described by an independent excitation, holons and spinons, with the latter forming the Fermi surface. Anderson (1991) proposed a scenario for understanding the observed linear temperature dependence of Pxx and the Hall carrier number nH= V/eRH in the normal state of copper oxide superconductors by distinguishing between the relaxation rates for carrier motion normal to the Fermi surface (transport scattering rate 1/Z'tr , a s measured in normal-state resistivity) and parallel to it (cyclotron relaxation rate 1/rH, as involved in Hall effect) if the transport properties are governed by two scattering times. According to his picture, one transport lifetime (Ttr) determines the longitudinal resistivity ( a ~ c~ "gtr), while both Ttr and TH determine the Hall conductivity (axy ~ rtrrH). The transport scattering rate 1/Ttr is linear in T while the transverse relaxation rate 1/TH, determined by spinon-spinon scattering, varies as T 2, like any other fermion-fermion interaction. Thus, a~y eC "Ct~rHeC 1/T 3. Since the Hall angle OH=tan-~(a~y/a~)oc 1/TH, the cotangent of the Hall angle cot OH =px~/RHH is then given by cot OH = AT 2 + B,

(8)

where A is related to the spinon bandwidth and B is a constant additive term due to magnetic impurity scattering. On the other hand, using the nearly antiferromagnetic Fermi-liquid description of planar quasiparticles, Stojkovic and Pines (1996) have shown that the anomalous temperature dependence of the resistivity and Hall coefficient and the quadratic temperature dependence of the Hall angle (eq. 8) can be the result of highly anisotropic scattering at different regions of the Fermi surface. Equation (8) was initially found to hold for YBa2Cu3_xZnxO7 ~ single crystals by Chien, Wang and Ong (1991b). It has subsequently been confirmed for a large number of systems including: T1 cuprates (Kubo and Manako 1992), YBa2Cu3 xCoxO7 - 6 (Carrington et al. 1992), YBa2Cu3_xFexO7 ~ (Lan et al. 1994), Yl-xPrxBa2Cu307_~ (Wu et al. 1992, Xiong, Xiao and Wu 1993, Almasan et al. 1994a), oxygen-deficient YBazCu3Oy (Xiong, Xiao and Wu 1993, Almasan et al. 1994b), Lai.85Sr0.15Cu1 xAxO4 (A=Fe, Co, Ni, Zn, Ga) (Gang, Peng and Cieplak 1992), and HgBa2CaCu206+~ (Harris et al. 1994). In sharp contrast to the complicated p~x(T) and nil(T) in different systems, eq. (8) seems to be universal in every system with different doping levels. Shown in fig. 17 is a plot of cot OH vs. T 2 calculated for a field of 6 T for Y1-xPrxBa2Cu307 ~ single crystals (Almasan et al. 1994a). Typically, in the optimally doped YBazCu307 ~ (Chien, Wang and Ong 1991b, Xiong, Xiao and Wu 1993), the

MAGNETORESISTANCE AND HALL EFFECT ,

800

~=

,

i

,

i

,

Yl.xPrxBa2Cu307_,~ +

600 . •

v o

275

o

o

o

[]

mu m

D° ~O • 0 0 o •° i I o m ID 013111 • ~DQO .t, o o m m o o o oO

400

X

o

...,,* , o=~,,oOO +°+.~++:VL,+ooO.._o°

200

o

0.0

- 0.08 [] 0.13

~,o8oO

+ 0 . 20.42 9"

0,51"

0

0

I

I

1

2

i

I

I

I

I

3

4

5

6

7

T2(104 K2) m 120

,

i

O~

~- 100 t-

i

Fig. 17. Cotangent of Hall angle cot(OH) vs. T2 ofY~ xPrxBa2Cu3Ov ~ single crystals with B = 6 T and B ]]c. From Almasan et al. (1994a).

80

..z 60

v O Bfor YvxPrBazCu 307,~~

~'o 4o < 20 o.

O

- 0- - A f°r YI-xPrxBa2Cu 307~

[3 Bfor YBa2Cu30y - I - - A forYBa2Cu30y

0 -20

,

0

I

20

i

I

i

40

D [] I

60 T(K)

[]

[] i

I

80

a

100

Fig. 18. Slope A and intercept B vs. To for Y ] - x P r x B a 2 C u 3 0 7 6 and YBa~Cu3Oy single crystals with B - 6 T and BI]c. From Almasan et al. (1994b).

T 2-dependence is observed in the temperature range 120-200 K. As the samples become more underdoped and the in-plane disorder increases (with increasing x), the range of the TZ-dependence moves towards higher temperatures. Figure 18 contains a plot of the values obtained for the slope A and the intercept B [determined by fitting the data of fig. 17 and similar data for oxygen-deficient YBazCu3Oy with eq. (8)] vs. Tc (Almasan et al. 1994b). Both A and B increase linearly with decreasing Tc in Y1 xPrxBa2Cu307 6, while in YBa2Cu3Oy only A changes with oxygen content while B stays constant. There are two interesting points to be made here: first, A is sensitive to carrier concentration, as predicted by Anderson (1991) and, second, the linear correlation between Tc and A revealed here implies a close relationship between the Hall effect and the superconducting mechanism. Xiong, Xiao and Wu (1993) reported a similar relationship between A and B, and Tc for epitaxial films of Y1-xPrxBa2Cu307 6, while Hall effect measurements of Yl-xPrxBazCu307 6 single crystals by Wu et al. (1992) revealed that only B increases with decreasing Tc while A remains constant within experimental resolution. Hall-effect measurements on YBa2Cu3 xZnxO? 6 single crystals performed by Chien, Wang and Ong (1991b) have shown that Zn doping, which introduces a local spin-½ impurity (Xiao et al. 1990) (bound spinon) into the CuO2 plane, only

276

C.C. ALMASAN and M.B. MAPLE

changes B proportional to x while maintaining a constant slope A. Therefore, the behavior of cot OH can be classified in three groups: (i) varying slope A and constant B, as in YBa2Cu3Oy; (ii) varying B and constant slope A, as in all Cu-site doped systems; and (iii) both varying A and B, typified by Y1 xPrxBa2Cu3OT_,~.

4. Summary and conclusions In this chapter, we have reviewed magnetoresistance and Hall effect measurements on high-Tc cuprate superconductors in both the normal and superconducting states that address several issues: for the normal state, the temperature dependence o f the Hall coefficient RH and the cotangent of the Hall angle cot OH; for the superconducting state, the temperature dependence o f the upper critical field Hc2 and irreversibility line H* as well as the anomalous sign change o f the Hall effect near To. Scaling behavior, dimensional cross-over phenomena, and relevant theoretical models were also discussed. These measurements have provided important information regarding the electronic structure and excitations that underly the normal and superconducting properties, fundamental superconducting-state parameters, as well as vortex pinning and dynamics of these remarkable materials. Magnetotransport studies, including those reviewed herein and many others outside the scope o f this article, have proven to be very useful in probing the physics o f the normal and superconducting states o f the high-To cuprate superconductors.

Acknowledgments We would like to thank Cassandra Rayt (UCSD) for preparing several figures for this article. This work was supported by the National Science Foundation under Grant Nos. DMR-9601839 and DMR-9801990 at Kent State University, and the U.S. Department o f Energy under Grant No. DE-FG03-86ER-45230 at University o f California at San Diego.

References Abrikosov, A.A., 1997, Phys. Rev. B 56, 5112. Almasan, C.C., and M.B. Maple, 1996, Phys. Rev. B 53, 2882. Almasan, C.C., C.L. Seaman, Y. Dalichaouch and M.B. Maple, 1991, Physica C 174, 93. Almasan, C.C., M.C. de Andrade, Y. Dalichaouch, J.J. Neumeier, C.L. Seaman, M.B. Maple, R.E Guertin, M.V. Km-icand J.C. Garland, 1992a,Phys. Rev. Lett. 69, 38t2. Almasan, C.C., S.H. Han, E.A. Early, B.Vv~Lee, C.L.

Seaman and M.B. Maple, 1992b, Phys. Rev. B 45, 1056. Almasan, C.C., S.H. Han, K. Yoshiara,M. Buchgeister, B.W. Lee, L.M. Paulius, D.A. Gajewski and M.B. Maple, 1994a, Physica B 199-200, 288. Almasan, C.C., S.H. Han, K. Yoshiara,M. Buchgeister, D.A. Gajewski, J. Herrmann, A.E Paulikas, G. Chtm, B.W. Veal and M.B. Maple, 1994b, Physica C 235 240(Part 5), 3119. Almasan, C.C., S.H. Han, K. Yoshiara,M. Buchgeister,

MAGNETORESISTANCE AND HALL EFFECT D.A. Gajewski, L.M. Paulius, J. Herrmann, M.B. Maple, A.P Paulikas, G. Chun and B.W. Veal, 1995, Phys. Rev. B 51, 3981. Anderson, P.W., 1990, Phys. Rev. Lett. 64, 1839. Anderson, EW., 1991, Phys. Rev. Lett. 67, 2092. Bardeen, J., and M.J. Stephen, 1965, Phys. Rev. A 140, 1197. Bergmann, G., 1979, Phys. Today 32, 25. Blatter, G., and B. Ivlev, 1993, Phys. Rev. Lett. 70, 2621. Budhani, R.C., S.H. Liou and Z.X. Cai, 1993, Phys. Rev. Lett. 71,621. Budhani, R.C., W.L. Holstein and M. Suenaga, 1994, Phys. Rev. Lett. 72, 566. Bud'ko, S.L., A.G. Gapotchenko, A.E. Luppov, E.A. Early, M.B. Maple and J.T. Markert, 1990, Physica C 168, 530. Carrington, A., A.E Mackenzie, C.T. Lin and J.R. Cooper, 1992, Phys. Rev. Lett. 69, 2855. Chien, T.R., T.W. Jing, N.E Ong and Z.Z. Wang, 1991a, Phys. Rev. Lett. 66, 3075. Chien, T.R., Z.Z. Wang and N.E Ong, 1991b, Phys. Rev. Lett. 67, 2088. Clayhold, J., N.E Ong, Z.Z. Wang, J.M. Tarascon and E Barboux, 1989, Phys. Rev. B 39, 7324. Clinton, T.W., A.V¢ Smith, Q. Li, J.L. Peng, R.L. Greene, C.J. Lobb, M. Eddy and C.C. Tsuei, 1995, Phys. Rev. B 52, R7046. Colino, J., M.A. Gonzalez, J.I. Martin, M. Velez, D. Oyola, E Prieto and J.L. Vincent, 1994, Phys. Rev. B 49, 3496. Cooper, YR., J.W. Loram and J.M. Wade, 1995, Phys. Rev. B 51, 6179. Dalichaouch, Y., B.W. Lee, C.L. Seaman, J.T. Markert and M.B. Maple, 1990, Phys. Rev. Lett. 64, 599. de Andrade, M.C., C.C. Almasan, Y. Dalichaouch and M.B. Maple, 1991, Physica C 184, 378. Dorsey, A.q2, 1992, Phys. Rev. B 46, 8376. Dorsey, A.T., and M.EA. Fisher, 1992, Phys. Rev. Lett. 68, 694. Feigel'man, M.V., V.B. Geshkenbein, A.I. Larkin and V.M. Vinokur, 1994, Physica C 235 240(Part 5), 3127. Ferrell, R.K., 1992, Phys. Rev. Lett. 68, 2524. Fert, A., and EM. Levy, 1987, Phys. Rev. B 36, 1907. Finnemore, D.K., R.N. Shelton, J.R. Clem, R.W. McCallum, H.C. Ku, R.E. McCarley, S.C. Chen, E Klavins and V Kogan, 1987, Plays. Rev. B 35, 5319. Fiory, A.T., and G.S. Grader, 1988, Phys. Rev. B 38, 9198.

277

Freimuth, A., C. Holm and M. Galffy, 1991, Phys. Rev. B 44, 10396. Galffy, K., and E. Zirngiebl, 1988, Solid State Cormaaun. 68, 929. Gang, X., X. Peng and M.Z. Cieplak, 1992, Phys. Rev. B 46, 8687. Geshkenbein, V.B., A.L Larkin, J.M. Harris, N.E Ong and Y.E Yah, 1994, Phys. Rev. Lett. 73, 609. Ginsberg, D.M., and R.L. Dudey, 1993, Physica A 200, 351. Ginsberg, D.M., and LT. Manson, 1995, Phys. Rev. B 51, 515. Goettee, J.D., J.S. Brooks, WJ. Skocpol, J.L. Smith, D.G. Rickel, B.L. Freeman, C.M. Fowler, RM. Mankiewich, E.I. De Obaldia and M.L. O'Malley, 1994, Physica B 194 196, 1805. Hagen, S.J., C.J. Lobb, R.L. Greene, M.G. Forrester and J.H. Kang, 1990a, Phys. Rev. B 41, 11630. Hagen, S.J., C.J. Lobb, R.L. Greene, M.G. Forrester and J. Talvacchio, 1990b, Phys. Rev. B 42, 6777. Hagen, S.J., C.J. Lobb, R.L. Greene and M. Eddy, 1991a, Phys. Rev. B 43, 6246. Hagen, S.J., C.J. Lobb, R.L. Greene and M. Eddy, 1991b, Physica C 185-189(Part 3), 1769. Hagen, S.L, A.W Smith, M. Rajeswari, J.L. Peng, Z.Y. Li, R.L. Greene, S.N. Mao, X.X. Xi, S. Bhattacharya, L. Qi and C.J. Lobb, 1993, Phys. Rev. B 47, 1064. Han, S.H., C.C. Almasan, M.C. de Andrade, Y. Dalichaouch and M.B. Maple, 1992, Phys. Rev. B 46, 14290. Hao, Z., J.R. Clem, M.W. McElfresh, L. Civale, A.P. Malozemoff and E Holtzberg, 1991, Phys. Rev. B 43, 2844. Harris, J.M., N.E Ong and Y.E Yan, 1993, Phys. Rev. Lett. 71, 1455. Harris, J.M., H. Wu, N.P. Ong, R.L. Meng and C.W Clm, 1994, Phys. Rev. B 50, 3246. Harris, J.M., N.P. Ong, P. Matt, R. Gagnon, L. Taillefer, T. Kimura and R. Kitazawa, 1995, Phys. Rev. B 51, 12053. Hidaka, Y., and M. Suzuki, 1989, Nature (UK) 338, 635. Hirsch, J.E., and E Marsiglio, 1991, Phys. Rev. B 43, 424. Iwasaki, H., Y. Dalichaouch, J.T. Markert, G. Nieva, C.L. Seaman and M.B. Maple, 1990, Physica C 169, 146. Iye, Y., S. Nakamura and T. Tamegai, 1989, Physica C 159, 616.

278

C.C. ALMASAN and M.B. MAPLE

Jia, Y.X., J.Z. Liu, A. Matsushita, M.D. Lan, P. Klavins and R.N. Shelton, 1992, Phys. Rev. B 46, 11745. Jia, Y.X., J.Z. Liu, M.D. Lan and R.N. Sheltun, 1993, Phys. Rev. B 47, 6043. Jin-Tae, K., J. Giapintzakis and D.M. Ginsberg, 1996, Phys. Rev. B 53, 5922. Jones, E.C., D.K. Christen and B.C. Sales, 1994, Phys. Rev. B 50, 7234. Kadowaki, K., and T. Mochiku, 1992, Physica C 195, 127. Kang, W.N., D.H. Kim, S.Y. Shim, J.H. Park, T.S. Hahn, S.S. Choi, W.C. Lee, J.D. Hettinger, K.E. Gray and B. Glagola, 1996, Phys. Rev. Lett. 76, 2993. Kang, WN., S.H. Yun, J.Z. Wu and D.H. Kim, 1997, Phys. Rev. B 55, 621. Kebin, L., Z. Yuheng and H. Adrian, 1996, Phys. Rev. B 53, 8608. Khomskii, D.I., and A. Freimuth, 1995, Phys. Rev. Lett. 75, 1384. Kopnin, N.B., 1996, Phys. Rev. B 54, 9475. Kopnin, N.B., and VE. Kravtsov, 1976a, JETP Lett. 23, 578. Kopnin, N.B., and VE. Kravtsov, 1976b, Pis'ma Zh. Eksp. Teor. Fiz. 23, 631. Kopnin, N.B., and A.V. Lopatin, 1995, Phys. Rev. B 51, 15291. Kopnin, N.B., B.I. Ivlev and V.A. Kalatsky, 1993, J. Low Temp. Phys. 90, 1. Kubo, Y., and T. Manako, 1992, Physica C 197, 378. Kunchur, M.N., D.K. Chi'isten, C.E. Klabtmde and J.M. Phillips, 1994, Phys. Rev. Lett. 72, 2259. Lan, M.D., J.Z. Liu, Y.X. Jia, Z. Lu and R.N. Shelton, 1994, Phys. Rev. B 49, 580. Larkin, A.I., and Yu.N. Ovchinnikov, 1995, Phys. Rev. B 51, 5965. Lindemann, E, 1910, Z. Phys. 11, 69. Lobb, C.J., T.W. Clinton, A.W. Smith, L. Wu, L. Qi, J.L. Peng, R.L. Greene, M. Eddy and C.C. Tsuei, 1994, Appl. Supercond. (UK) 2, 631. Luo, J., T.P Orlando, J.M. Graybeal, X.D. Wu and R. Muenchausen, 1992, Phys. Rev. Lett. 68, 690. Mackenzie, A.R, S.R. Julian, G.G. Lonzarich, A. Carrington, S.D. Hughes, R.S. Liu and D.C. Sinclair, 1993, Phys. Rev. Lett. 71, 1238. Maki, K., 1966, Phys. Rev. 148, 362. Malozemoff, A.R, 1989, Macroscopic magnetic properties of high temperature superconductors, in: Physical Properties of High Temperature Superconductors, Vol. 1, ed. D.M. Ginsberg (World Scientific, Singapore) p. 71.

Malozemoff, A.P., T.K. Worthington, Y. Yeshurtm, E Holtzberg and P.H. Kes, 1988, Phys. Rev. B 38, 7203. Manson, J.T., J. Giapintzakis, A.I. Schegolev and D.M. Ginsberg, 1996, Phys. Rev. B 54, 10155. Maple, M.B., Y. Dalichaouch, J.M, Ferreira, R.R. Hake, B.W Lee, J.J. Neumeier, M.S. Torikachvili, K.N. Yang, H. Zhou, R.R Guertin and M.V. Kuric, 1987, Physica B&C 148B+C, 155. Maple, M.B., B.W. Lee, J.J. Neumeier, G. Nieva, L.M. Paulius and C.L. Seaman, 1992, J. Alloys & Compotmds 181, 135. Maple, M.B., C.C. Almasan, C.L. Seaman, S.H. Han, K. Yoshiara, M. Buchgeister, L.M. Paulius, B.W. Lee, D.A. Gajewski, R.E Jardim, C.R. Fincher Jr, G.B. Blanchet and R.R Guertin, 1994, J. Supercond. 7, 97. Matsuda, Y., T. Nagaoka, G. Suzuki, K. Kumagai, M. Suzuki, M. Machida, M. Sera, M. Hiroi and N. Kobayashi, 1995, Phys. Rev. B 52, R15749. Nakao, K., N. Miura, K, Tatsuhara, H. Takeya and H. Takei, 1989, Phys. Rev. Lett. 63, 97. Neiman, R.L., J. Giapintzakis, D.M. Ginsberg and J.M. Mochel, 1995, J. Supercond. 8, 383. Niessen, A.K., EA. Staas and C.H. Weijsenfeld, 1967, Phys. Lett. 25A, 33. Noto, K., S. Shinzawa and Y. Muto, 1976, Solid State Commun. 18, 1081. Nozi+res, P, and WE Vinen, 1966, Philos. Mag. 14, 667. Oshima, K., H. Urayama, H. Yamochi and G. Saito, 1988, Physica C 153, 1148. Osofsky, M.S., R.J. Soulen Jr, S.A. Wolf, J.M. Broto, H. Rakoto, J.C. Ousset, G. Coffe, S. Askenazy, R Pari, I. Bozovic, J.N. Eckstein and G.E Virshup, 1993, Phys. Rev. Lett. 71, 2315. Ovchinnikov, Y.K., and V.Z. Kresin, 1995, Phys. Rev. B 52, 3075. Ovchinnikov, Y.K., and V.Z. Kresin, 1996, Phys. Rev. B 54, 1251. Paulius, L.M., C.C. Almasan and M.B. Maple, 1993, Phys. Rev. B 47, 11627. Paulius, L.M., B.W Lee, M.B. Maple and RK. Tsai, 1994, Physica C 230, 255. Qiu, X.G., G. Jakob, VV. Moshchalkov, Y. Bruynseraede and H. Adrian, 1995, Phys. Rev. B 52, 12994. Rhyne, J.J., 1968, Phys. Rev. 172, 523. Rice, J.R, J. Giapintzakis, D.M. Ginsberg and J.M. Mochel, 1991, Phys. Rev. B 44, 10158. Rice, J.R, N. Rigakis, D.M. Ginsberg and J.M. Mochel, 1992, Phys. Rev. B 46, 11050.

MAGNETORESISTANCE AND HALL EFFECT Samoilov, A.V., 1993, Phys. Rev. Lett. 71, 617. Samoilov, A.V., Z.G. Ivanov and L.G. Johansson, 1994, Phys. Rev. B 49, 3667. Schilling, A., R. Jin, J.D. Guo and H.R. Ott, 1993, Phys. Rev. Lett. 71, 1899. Seidler, G.T., T.E Rosenbaum, D.L. Heinz, J.W. Downey, A.E Paulikas and B.W. Veal, 1991, Physica C 183, 333. Stojkovic, B.P., and D. Pines, 1996, Phys. Rev. Lett. 76, 811. Sttmarlin, I.W., S. Skanthakumar, J.W. Lynn, J.L. Peng, Z.Y. Li, W Jiang and R.L. Greene, 1992, Phys. Rev. Lett. 68, 2228. Suzuki, M., and M. Hikita, 1990, Phys. Rev. B 41, 9566. Tinkham, M., 1988, Phys. Rev. Lett. 61, 1658. Tinkham, M., ed., 1996, Introduction to Superconductivity (McGraw-Hill, New York). Ullah, S., and A.T. Dorsey, 1991, Phys. Rev. B 44, 262. Van Otterlo, A., M.V. Feigel'man, V.B. Geshkenbein and G. Blatter, 1995, Phys. Rev. Lett. 75, 3736. Vinokur, V.M., V.B. Geshkenbein, M.V. Feigel'man and G. Blatter, 1993, Phys. Rev. Lett. 71, 1242.

279

Walker, D.J.C., O. Laborde, A.P. Mackenzie, S.R. Julian, A. Carrington, LW. Loram and J.R. Cooper, 1995, Phys. Rev. B 51, 9375. Wang, L.M., H.C. Yang and H.E. Horng, 1997, Phys. Rev. Lett. 78, 527. Wang, Z.B., D. Jinming and C.S. Ting, 1994, Phys. Rev. Lett. 72, 3875. Wang, Z.D., and C.S. Ting, 1991, Phys. Rev. Lett. 67, 3618. Welp, U., W.K. Kwok, G.W. Crabtree, K.G. Vandervoort and J.Z. Liu, 1989, Phys. Rev. Lett. 62, 1908. Welp, U., S. Fleshier, W.K. Kwok, R.A. Klemm, VM. Vinokur, J. Downey, B. Veal and G.W. Crabtree, 1991, Phys. Rev. Lett. 67, 3180. Werthamer, N.R., E. Helfand and P.C. Hohenberg, 1966, Phys. Rev. 147, 295. Wu, J., J.L. Peng, S.J. Hagen and R.L. Greene, 1992, Phys. Rev. B 46, 8694. Xiao, G., M.Z. Cieplak, LQ. Xiao and C.L. Chien, 1990, Phys. Rev. B 42, 8752. Xiong, R, G. Xiao and X.D. Wu, 1993, Phys. Rev. B 47, 5516. Zavaritsky, N.V, A.V. Samoilov and A.A. Ym'gens, 1991, Physica C 180, 417.

Handbook on the Physics and Chemistry o f Rare Earths Vol. 31 edited by K.A. Gschneidner, Jr., L. Eyring and M.B. Maple © 2001 Elsevier Science B. g All rights reserved

Chapter 198

NEUTRON SCATTERING STUDIES O F S P I N F L U C T U A T I O N S IN HIGH-TEMPERATURE SUPERCONDUCTORS T.E. Mason* University of Toronto, Department of Physics, 60 St. George St., Toronto, ON M5S 1A 7, Canada, and Spallation Neutron Source, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

Contents List of symbols and acronyms Introduction 1. Neutron scattering cross-section 2. Insulating cuprates 2.1. Antiferromagnetism in La2CuQ 2.1.1. Spin waves 2.1.2. Paramagnetic critical scattering 2.2. Lightly doped La 2 x(Sr,Ba),CuO4 2.3. Antiferromagnetism in YBa2Cu306 ~x 3. Superconducting La 2 xSrxCuO4 3.1. Normal-state spin fluctuations 3.1.1. Low-energy incommensurate response

281 282 283 284 284 286 290 292 294 297 297

3.1.2. Comparison of high-energy spin dynamics with La2CuQ 302 3.2. Superconductivity and the incommensurate spin fluctuations 304 4. Superconducting YBa2Cu307 x 305 4.1. Low-energy incommensurate response 307 4.2. The (&co) resonance 309 5. Conclusions 311 Acknowledgements 312 References 312

297

List o f symbols and acronyms c

f(Q) hw Jll

Ji k (k')

Q s°~(Q,~o) Zo, Z x

spin-wave velocity magnetic form factor neutron energy transfer nearest-neighbour antiferromagnet exchange constant interlayer exchange constant incident (scattered) neutron wave vector momentum transfer magnetic scattering function

F ~c /~ (m~) d 2a dg2 dE X"(Q,o~)

quantum renormalization parameters for exchange constant and magnetic response

inverse lifetime inverse correlation length Bohr magneton correlation length antiferromagnetic zone center for CuO 2 planes partial differential cross-section imaginary part of the generalized magnetic susceptibility

* Present Address: Spallation Neutron Source, Oak Ridge National Laboratory, Oak Ridge, TN, 37831, USA. 281

282

T.E. MASON

Introduction

Soon after the discovery of high-temperature superconductors in 1986 it was realised that superconductivity occurred in close proximity to antiferromagnetic ordering of the spin-½ Cu 2+ moments present in the CuO2~planes which typify the high-Tc superconductors. The dynamics of these spins, both in the undoped, insulating parent compounds and in the doped, metallic, superconductors represent an intriguing topic of study, in the former case as an example of a low-dimensional quantum antiferromagnet and in the latter case as a strongly correlated metal. This chapter deals with neutron scattering studies of the spin fluctuations in the CuO2 planes; as such, the rare earths that form an important constituent of almost all high-Tc materials are viewed as somewhat passive spectators. The fact that many candidate theories of high-temperature superconductivity include spin fluctuations as a central feature of the mechanism for superconductivity increases the importance of obtaining a detailed picture of the energy, momentum, temperature, and doping dependence of the magnetic dynamics of the cuprates. Furthermore, many of the anomalous normal-state properties of the doped cuprates are likely related to the proximity of antiferromagnetism in the phase diagram and the manifestation of this instability in the excitations. Even in the absence of a magnetic mechanism for hightemperature superconductivity the spin fluctuations are an important degree of freedom of the quasiparticles which form the superconducting condensate and therefore a probe of the nature of that transition. Information about spin excitations may be inferred indirectly from bulk measurements such as magnetic susceptibility or heat capacity (see, for example, ch. 200). Nuclear magnetic resonance provides a more microscopic probe of the low-energy spin fluctuations and can provide some information about the momentum dependence when several different nuclear sites are compared. Raman spectroscopy (ch. 203) can measure the twomagnon cross-section and therefore is sensitive to the energy scale over which magnetic excitations occur and its evolution with doping or temperature. The most detailed probe of spin dynamics is inelastic neutron scattering (for a recent overview see Aeppli et al. 1997a), the topic of the present chapter. Because of its magnetic moment the neutron is sensitive to magnetic moments in solids. It is a weakly interacting, non-perturbing probe with a well-understood cross-section that is directly proportional to the static and dynamic spin correlation functions. Elastic neutron scattering can be used to determine the arrangement of spins in an ordered magnetic state (see, for example, ch. 199 describing the 4f magnetic ordering that occurs in many of the rare-earth cuprate superconductors). Inelastic neutron scattering can probe crystal field excitations (ch. 194 of volume 30 of this Handbook) as well as the collective excitations of spins, ordered or disordered. While the neutron is in many respects an ideal probe of magnetism in solids it has the drawback that neutron sources are relatively weak (even the highest flux neutron source produces fewer neutrons than a bench-top Cu-Kc~ X-ray tube). As a result, very large single crystals (of order 1 cm 3 or bigger) are required to measure the weak cross-section inherent in any inelastic experiment. For the high-To materials this situation is made more difficult by the small spin quantum number (½) and the large energy scale over which spin

NEUTRON SCATTERINGSTUDIES OF SPIN FLUCTUATIONSIN HTSC

283

fluctuations exist (hundreds of meV). These limitations mean that only two systems have been studied with inelastic neutron scattering in any detail, La2_x(Sr, Ba)xCuO4 (214) and YBa2Cu307-x (123). There have been some studies of spin fluctuations in other high-To systems [for example Pr and Nd 214; see, e.g., Matsuda et al. (1993) and Ivanov et al. (1995)]; however, the antiferromagnetic spin waves are not qualitatively different from those observed in La 214 and the experimental difficulties of small crystals have precluded any detailed study of superconducting samples. Neutron scattering studies of 4f antiferromagnetic ordering in the rare-earth high-Tc superconductors are discussed in ch. 199 of this volume. Following a brief summary of the neutron scattering crosssection the 214 and 123 systems are discussed, in turn, with the spin fluctuations in the superconductors as the main emphasis.

1. Neutron scattering cross-section Because of its magnetic moment the neutron can couple to moments in solids via the dipolar force. The energies and wavelengths of thermal and cold neutrons are well matched to the energy and length scales of most condensed-matter systems. In the case of high-temperature superconductors, the characteristic energy of the spin fluctuations is determined by the superexchange interaction between nearest-neighbour copper spins. This is typically ~150meV so a complete characterisation of the magnetic dynamics requires the use of higher-energy sources such as those found at spallation neutron sources or a reactor hot source. We will briefly review the formalism that describes magnetic neutron scattering. For a detailed treatment of the neutron scattering cross-section, there are some excellent texts which can serve as an introduction (Squires 1978) or a more comprehensive exposition (Lovesey 1984). The partial differential cross-section for magnetic neutron scattering, which measures the probability of scattering per solid angle per unit energy, is d2• dE2 dE

UN

(1) a~

where k (U) is the incident (scattered) neutron wavevector, N is the number of moments, yr0=5.391 fm is the magnetic scattering length, f(Q) is the magnetic form factor (analogous to the electronic form factor appearing in the X-ray scattering cross-section), Q is the momentum transfer, o) is the energy transfer, and the sunamation runs over the Cartesian directions. Sa~(Q,o0) is the magnetic scattering function which is proportional to the space and time Fourier transform of the spin-spin correlation function. If the incident and scattered neutron energies are the same (elastic scattering) then the correlations at infinite time are being probed and, in a magnetically ordered material, the scattering function will contain delta functions at the wavevectors corresponding to magnetic Bragg reflections. The ( 6 a ~ - QaQ/3) term in the cross-section means that neutrons probe the components of spin perpendicular to the momentum transfer, Q. A

284

T.E. MASON

If there is no analysis of the scattered neutron energy then (within the static approximation) the measured intensity is proportional to the Fourier transform of the instantaneous correlation function which is essentially a snapshot of the spin correlations in reciprocal space. At non-zero energy transfer, the spin dynamics of the system under study are probed. In a magnetically ordered,system of localised spins, the elementary magnetic excitations are spin waves. The fluctuation-dissipation theorem relates the correlations to absorption; in other words, the scattering function is proportional to the imaginary part of a generalised (Q and c0 dependent) susceptibility, z'(Q,o)). In the zero-frequency, zero-wavevector limit, the real part of the generalised susceptibility is the usual DC susceptibility measured by magnetisation. In a metal, the elementary excitations are electron-hole pairs. Since it is possible to excite an electron-hole pair by promoting a quasiparticle from below the Fermi surface to above the Fermi surface, and at the same time flip its spin, neutrons can be used to probe the low-energy excitations of a metal. The generalised susceptibility (for a non-interacting metal) is just the Lindhard susceptibility that can be calculated from the band structure.

2. Insulating cuprates 2.1. Antiferromagnetism in La2Cu04 The high-Tc superconductors are complex materials displaying, in a single sample, diverse phenomena that have been active topics of research in condensed matter physics for the last thirty years. They are oxides of copper having perovskite crystal structures, generally tetragonal at high temperatures and orthorhombic at low temperatures. The importance of magnetism in the undoped, insulating, parent compounds was confirmed quite soon after the initial discovery of superconductivity in La2-xBaxCuO4 when Vaknin et al. (1987) showed, using neutron powder diffraction, that anomalies seen in the magnetic susceptibility of La2CuO4_y (Johnston et al. 1987) were due to a transition to a three-dimensional, long-range, ordered, antiferromagnetic state. The Cu 2+ ions of the CuO2 sheets that typify this class of materials are coupled via an antiferromagnetic superexchange interaction that is considerably stronger between nearest-neighbour copper spins in the nearly square CuO2 planes than perpendicular to them. This leads to effectively two-dimensional magnetic behaviour; the transition to long-range 3D order occurs due to the weak interplane interaction which is not large enough to affect the spin dynamics very significantly. Figure 1 shows the weak superlattice reflection observed below TN by Vaknin et al. (1987) in their unpolarized neutron powder diffraction measurements. The lower panel shows how this peak, which corresponds to the unit cell doubling of a simple + - arrangement of nearest-neighbour spins on a square lattice, develops below TN (the actual value of TN depends on the oxygen doping, y - it is 220 K for the sample used in fig. 1). The fact that this reflection is magnetic in origin was subsequently confirmed by

NEUTRON SCATTERING STUDIES OF SPIN FLUCTUATIONS IN HTSC

3500 C

I

3000

T = 15 K

I

285

(a)

x,~z ~ A~

L

A

2500

A

d~

H

z 2000

W

T = 300 K

Z H

o

1500

,

22

0

I

24 SCATTERING

/ I

,

i

28

26 ANGLE

(20, BEG)

600

(b) c

400

c_

200

~

I---

o

tt '

z -200 o

o

'

'2

TEMPERATURE

o

(K)

"

30O

Fig. l. Antiferromagnetic ordering observed in La2CuO4 y by unpolarized powder neutron diffraction. The top panel shows the (100) peak which appears below T N (=220K)0 and the bottom panel shows the temperature dependence of the Bragg intensity. From Vaknin et al. (1987).

polarised neutron measurements (Mitsuda et al. 1987). The ordered moment corresponds to 0.4/~B/Cu. The ordering wavevector, (100) in orthorhombic notation or (Jv,~) in units of the dimensionless 2D reciprocal lattice, corresponds to the Fermi surface nesting instability of the 2D square lattice from which the low-temperature structure is derived. A variety of experimental techniques have been used to establish the behaviour of La2_x(Sr,Ba)xCuO4 as holes are introduced into the CuO2 planes by replacing La with Sr or Ba. Figure 2 summarises the results for Sr doping. Initially the effect of doping is to reduce TN, which vanishes beyond x ,~ 0.03. There is a region of spin-glass behaviour before. For doping levels in the range 0.07-0.25, superconductivity occurs with a maximum Tc of ~ 3 9 K for x=0.15, Because the 3d 9 ionic configuration of Cu 2+ has an effective spin -1 ground state with no orbital angular momentum, La2CuO4 is an excellent realisation of a 2D square lattice Heisenberg antiferromagnet that corresponds to a low-dimensional quantum Hamiltonian which, before the advent of high Tc, was not understood. There is a small spin anisotropy due to antisymmetric exchange; an indication of the extent to which this causes deviation from Heisenberg symmetry can be seen in the small canting (~0.003 rad) of the spins out of the CuO2 plane (Kastner et al. 1988). Because it is such an excellent realisation of a model quantum magnet, experiments on

286

T.E. MASON

Lo2-xSrxCu04 PhoseDiogrom 6OO

\

~%o

rE

"%.0 \ \

,./ 4 0 0

TRA

-

GONA L

\

300

E Ic ~

200

%

t

\

\

100

Ll~LOC. 0

.

.

.

.

.

.

.

...h~-.-~-'=~ .

.

.

•

l

•

T%. x~.

0.1 0.2 0.3 Sr concentration x

o

0.4

Fig. 2. Phase diagram for La2_xSrxCu04 showing the phase boundaries for the tetragona~ orthorhombic, paramagnetic-antiferromagnetic, and normal-superconducting phase transitions as a fimction of temperature and doping. From Keimer et al. (1992).

L a 2 C u O 4 have been crucial as a test of theories that purport to explain this system. L a z C u O 4 is now probably the most completely characterised model magnetic system in

existence. The spin dynamics and critical scattering have been quantitatively measured and understood over an unprecedented range of energy and temperature. 2.1.1. Spin w a u e s

The lamellar structure of the high-T~ oxides suggests that the magnetic properties of the spin-2t- moments in the CuO2 planes will be quasi-two-dimensional, as is the case for isostructural magnetic materials such as K2NiF4 and K2MnF4 (Birgeneau, Als-Nielsen and Shirane 1977). The appropriate magnetic Hamiltonian for such a system is that of weakly coupled planes of Heisenberg spins:

H = ~ JiiSi. Sj + Z J±Si" Sj,. ij

(2)

fi'

The first term represents the coupling between nearest-neighbour Cu spins in the same CuO2 plane. This is the dominant interaction and, neglecting the inter-layer coupling described by the second term in eq. (2), yields the Hamiltonian of a (nearly) square lattice Heisenberg antiferromagnet. One notable feature of La2CuO4 is that it is the first good example of a Heisenberg (i.e., isotropic) spin- 1 system in two dimensions studied in any detail. Unlike systems with an Ising anisotropy, there is no phase transition to a long-range

NEUTRON SCATTERINGSTUDIES OF SPIN FLUCTUATIONSIN HTSC

287

ordered state at finite temperature expected for Heisenberg spins. The ordering below TN that occurs in La2CuO4 and other high-To compounds is driven by the weak interplane interaction that eventually leads to 3D ordering. Well below TN, in the zero-temperature limit, the excitations out of the antiferromagnetically ordered ground state of eq. (2) are spin waves. Neglecting interlayer coupling, conventional spin-wave theory in the classical (large-S) limit predicts a dynamic susceptibility (as measured by inelastic neutron scattering) of the form

x"(Q,~o) = zzJrg2#~S

l1+- yy(Q)'~ ( Q ) ] 1/2 6(hfo -4- boo(Q)),

(3)

where

hoo(Q) = 2Jii [1 - y2(Q)]

(4)

and y(Q) = cos(Jrh)cos(Jr/). Equation (3) includes a parameter, Zz, which reflects the renormalization of the magnetic response compared to the classical theory due to the fact that S is not large and the N6el state is not a good approximation of the ground state. There is also a renormalization of the energy scale of the spin waves such that the exchange constant deduced from measurements of the spin-wave dispersion relation, J~, is related to that appearing in eq. (2) via Jl~ = ZcJ. In the case of the S = ½ squarelattice antiferromagnet Singh (1989) and Igarashi (1992a,b) have estimated Zc = l. 18 and Z X = 0.51 based on a 1/S expansion. Since neutron scattering measures only the effective dispersion, it cannot be used to determine Zc. However, by placing intensity measurements on an absolute scale, Z z can be measured directly. Initial attempts to measure the spin waves in La2CuO4 were frustrated by the very large value of the exchange constant JN, a consequence of the strong superexchange path between Cu 2+ ions via the planar O. This implies that the slope of the dispersion relation, or spin-wave velocity, is very steep compared to magnetic systems typically studied using thermal neutron, triple-axis spectroscopy. For energy transfers below -50meV, which are easily accessed by a reactor-based, thermal neutron instrument the two counterpropagating spin-wave modes emerging from an antiferromagnetic reciprocal lattice point are not sufficiently well separated in momentum space to be resolved due to the finite instrumental resolution. For this reason, measurements such as those shown in fig. 3 were only able to place a lower bound on the spin-wave velocity (and, hence, exchange constant Jll) of > 4 0 0 m e V A (Shirane et al. 1987, Endoh et al. 1988, Yamada et al. 1989). In order to access energy transfers large enough to separately resolve the spin waves, an intense source of high-energy neutrons, such as that available at the "hot source" at the Institut Laue-Langevin, is required. A "hot source" consists of a block of graphite that is heated to a temperature much above the ambient temperature of the heavy water that acts a moderator for most of the neutrons. The neutrons that come to thermal equilibrium within the hot source have a Maxwellian distribution characterised by a temperature of 2000 K, as opposed to 300 K for the reactors thermalized in the

288

T.E. M A S O N L a 2 C u 0 4 ., 14.7Ef. 40L40"40240 '. ( h, 0.24)

=

E

[ [ 244 K

120K

5.7K

12mev

5O

j-% aoo ~ J

0

cJ -....

6mev

" 50

n

o

t \

Ji °

/I

o~

.,--~" 0

'

~

)';i %o

0 °9"o °~o_oo i

,;o~,

'

~_100 r

;

? i I

o~ 'o

/

ff 0.9

1.0

1.1

n 0.9

i 1.0

h( r. I,u.)

i 1.1

II

'

0.9

1.0

1.1

Fig. 3. Thermal triple-axis constant-energy-transfer scans through the antiferromagnetic zone center, (100), at selected temperatures below TN, for La2CuO 4. The dashed lines show fits to conventional spin wave theory convolved with the spectrometer resolution. In all cases, the observed peaks are unable to resolve the counterpropagating spin wave modes. This type of measurement can only place a lower bound ( 6 5 0 m e V A ) on the spin-wave velocity. From Yamada et al. (1989).

heavy water. By allowing measurements to higher energy transfers the spin-wave velocity can be unambiguously determined from the location of resolved spin-wave peaks in constant-energy scans (see fig. 4) (Aeppli et al. 1989, Hayden et al. 1990). In order to fully determine the parameters of the magnetic Hamiltonian that describes the excitations in La2CuO4, it is necessary to measure spin-wave excitations throughout the Brillouin zone. This can only be done using the special characteristics of a spallation neutron source that has a spectrum of undermoderated neutrons extending beyond the thermal and hot range. Figure 5 shows the zone-boundary magnon for La2CuO4 measured on the HET spectrometer at ISIS along with the full dispersion relation obtained from the HET and IN1 measurements (Hayden et al. 1991b). In addition to determining the exchange constant, J~ = 153+4meV (corresponding to a J of 135 meV), these measurements, via the ratio of the zone-boundary magnon energy to the spin-wave velocity (850 meV A), place an upper bound on the next-nearest-neighbour exchange constant at 9 meV In addition, the width of observed zone-boundary magnon places a lower bound on the spinwave lifetime of 9.3 x 10 13 s. More recent measurements of the spin waves in La2CuO4 over the full Brillouin zone have been placed on an absolute scale (Hayden et al. 1996a, 1998; see also sect. 4.1.2), allowing a determination of the quantum renormalization of the magnetic response. Quantitative analysis of the intensity of the spin-wave response yields an estimate of Z x = 0.394-0.1, in reasonable agreement with the calculation of Igarashi (1992a,b). Taken together, the inelastic neutron scattering studies of La2CuO4 provide a quantitative verification of the applicability of renormalized spin-wave theory and a determination of the parameters of the magnetic Hamiltonian (eqs. 2-4) that describes this material.

NEUTRON SCATTERING STUDIES OF SPIN FLUCTUATIONS IN HTSC

La2Cu04 T=290 K

,'~,,;,;e~' I ~'=(3°°~

'-'

-12200

~

l

;,++

E

E

........ a,,

2000 ¢o

++ I,.oo

,

'ho)=lO0 m e V

289

~,l,J.

1400

,5 E 1200

b-,

z~

1ooo

E i

i

q

;

;

I1o)=50 meV

~

',

'

~

:

~

:

~' ~

~

1

..~,0m

400

¢'4

', ~ | 4 0 0

~

i 2°°

cO

3013

'~ 2013 -

o18' '-0.3' ' ' ~ ..... 0.3

o'.6

q(k')

(a)

La2CuO 4

Fig. 4. Constant-energy scans through the counter-propagating spin wave modes of La2CuO 4. Using higher-energy neutrons available at a "hot source", it is possible to measure the spin waves at frequencies where the two branches emerging from (100) and equivalent positions are separable, allowing a determination of the spin wave velocity. The dashed line in the upper scan shows the resolution-limited response, while the solid curves are the results of fits to a resolutionconvolved spin wave spectrum with a velocity of 850meV,~. From Hayden et al. (1990).

(b)

,:,~

200

~.'-iAF 0

IO0

0 200

300

400

1~= (meV)

0

0.2

0.4

0.6

q II(A-~

Fig. 5. Zone-boundary magnon and spin wave dispersion of La~CuO4. Spallation neutron sources have sufficient undermoderated epithermal neutrons to permit measurements with incident neutron energies up to 2 eV, making it possible to access high enough energy transfer to observe the zone-boundary magnon mode in La2CuO 4 (panel a) at 312meV The time-of-flight data obtained on HET at ISIS (solid circles) combined with the triple-axis measurements shown in fig. 4 (open circles) were used to determine the full dispersion relation for the spin waves (panel b). From Hayden et al. (1991b).

290

T.E. MASON

The spin dynamics of La2CuO4 are not dramatically affected by the N6el transition, a reflection of the fact that there are quite long-range 2D correlations in the paramagnetic state (see below). The three-dimensional ordering is driven by the weak inter-layer coupling and has little effect on the intra-layer dynamics for energies comparable to J* beyond introducing a modest 3.6+2.9meV damping (at 320K, Hayden et al. 1990). At low (<5 meV) energies, there is a manifestation of the transition to the paramagnetic state in the form of an additional, quasielastic contribution to the cross-section which has a width of 1.5+0.4meV, in good agreement with expectations (Grempel 1988). This contribution is due to spin-wave interactions and is also consistent with numerical simulations (Tyc et al. 1989). For energies hco >~hc~ -1, renormalized spin-wave theory gives an excellent account of the dynamics, even in the paramagnetic state.

2.1.2. Paramagnetic critical scattering The properties of a magnetic system close to an instability such as an antiferromagnetic ordering transition are characterised by divergences in various physical properties. Neutron scattering can access several of the important parameters through the energyintegrated cross-section, or critical scattering, which is proportional to the Fourier transform of the instantaneous spin-spin correlation function (for a detailed discussion see Collins 1989). Of course, one can numerically integrate the measured inelastic crosssections described in sect. 2.1.1; however, this is not as efficient as performing the energy integration experimentally by detecting scattered neutrons without energy analysis. This is accomplished on a triple-axis spectrometer by operating in the two-axis mode, without an analyser crystal. Two-axis mode will measure the desired energy-integrated cross-section only if the variation in momentum transfer for varying final energy can be neglected, the so-called static approximation. For a two-dimensional system, this can be assured by choosing the experimental geometry such that the final wavevector of the scattered neutrons is perpendicular to the planes (in this case the cross-section is integrated at constant in-plane wavevector; Birgeneau et al. 1977). This geometry is typically employed in studies of the critical scattering in high-To cuprates. Provided the incident neutron energy is high enough to span the range of important energy transfers, the instantaneous spin correlations are probed. Proper integration is an important consideration for La2CuO4 given the large exchange energy; calculations based on the cross-section expected for pure La2CuO4 suggest there is not a problem in this case or for lightly doped cuprates using conventional techniques (Keimer et al. 1992). The measured neutron intensity is typically a Lorentzian centered on the ordering wavevector, (100) or (~,~), with a height proportional to the staggered susceptibility and width proportional to the inverse correlation length. Figure 6 shows results obtained above TN for La2CuO4 by Keimer et al. (1992). Aside from a region of temperature extremely close to the 3D phase transition, the antiferromagnetic correlations in LazCuO4 are essentially two-dimensional in character due to the very large anisotropy in the magnetic interactions in the Hamiltonian (eq. 2). The Q-dependent static susceptibility of the 2D S = ½ Heisenberg antiferromagnet has

NEUTRON SCATTERING STUDIES OF SPIN FLUCTUATIONSIN HTSC

291

La2Cu04.0 (TN=325K)

"325 K ~ '

1000

"

800 600 400 om

E

400 K

1000 800

C

= 600

0 ,.%o

U'b ¢...

~00

500 K

1000

I

,

I

,

800 G00 400

o

~

O

0.0

0.95 1.00 1.05 1.10 H (r.l.u.)

Fig. 6. Paramagnetic critical scattering in La2CuO4. Energy-integrating scans across the two-dimensional rod of scattering measure the Fourier transform of the instantaneous spin spin correlation function in La2CuO4 above TN. The peak width (corrected for finite instrumental resolution) is the inverse correlation length for antiferromagnetic spin fluctuations. The dashed lines show the experimental resolution function, and the solid lines are the results of fits to a Lorentzian lineshape convolvedwith the resolution for three different temperatures. From Keimer et al. (1992).

been treated in detail by Chakravarty et al. (1989) and Hasenfratz and Niedermayer (1991). In the regime governed by the antiferromagnetic instability (the renormalized classical region), the temperature dependence o f the correlation length is determined by the exchange interaction according to the expression

a

- 0.493e llsd/r

1 - 0.43

+O

.

(5)

Figure 7 shows a comparison between this prediction and the measured inverse correlation length for La2CuO4 between TN and 8 0 0 K (Birgeneau et al. 1995). The value o f J extracted from the low-temperature spin-wave measurements is used in the comparison so there are no adjustable parameters. There is very good agreement over the range o f temperatures probed with the renormalized classical predictions, perhaps to a greater extent than might be expected based on the range over which the theory is likely to be valid. Additional studies on another realisation o f the spin -1 2D Heisenberg

292

T.E. MASON

0.10

La2Cu04.o Inverse Magnetic Correlation Length Renormalized Classical o

0.08

115£i

I

_

90El o 41Ei • 14.7Ei

0.06 "-'z =.,.

0.04

0.02 TN

0 300

i

400

i

500

600

700

800

T (K)

900

Fig. 7. Inverse correlation length exn'acted from fits to two-axis, energy-integrating scans through the antiferromagnetic position (see fig. 6). The line is the result of a fit to the renormalized classical theory described in the text. From Birgeneau et al. (1995).

antiferromagnet with a smaller J have covered an even wider range of T/J and found similar good agreement (Greven et al. 1994, 1995). There is also no evidence for a crossover to a quantum-critical regime as suggested based on the N M R measurements of Imai et al. (1993). Somewhat perplexingly, given the excellent agreement for the correlation length, the temperature dependence of the staggered susceptibility (peak intensity) is described by S(0) ~ ~2, not the expected S(0) ~ ~2 T 2. 2.2. Lightly doped La2_x(Sr,Ba)xCu04 Doping La2CuO4 by replacing La with Sr or Ba introduces holes in the CuO2 planes, eventually leading to an insulator-metal transition and a superconducting ground state. In addition to the changes in the charge transport, the low-energy magnetic properties are modified by these impurities. The N6el temperature is rapidly suppressed by introduction of these impurities due to the frustration associated with a hole that substitutes a ferromagnetic bond for an antiferromagnetic one. Beyond x ~ 0.015 there is no longer a long-range ordered anti ferromagnetic state at low temperatures. The range 0.015 ~<x ~< 0.05 is characterised by carrier localization below ~ 1 0 0 K and twodimensional spin glass freezing (Harshman et al. 1988, Sternlieb et al. 1990). Despite the absence of long-range order at these intermediate doping levels, there are still strong antiferromagnetic correlations as evidenced by peaks in the energy-integrated neutron cross-sections at the same position, ( & ~ ) , as in undoped La2CnO4 (Keimer et al. 1992). Qualitatively, the results are similar to those described in sect. 2.1.2 for the paramagnetic state of La2CuO4, although the inverse correlation length does not vanish:

NEUTRON SCATTERINGSTUDIES OF SPIN FLUCTUATIONSIN HTSC

0.06

293

Inverse Magnetic Correlation Length •

I

o La2Cu04.0

I

a Lal.gaSro.ozCuO 4 v Lal.97Sro.o3CuO 4

0.05 0.04 ~'0.03

0.02 i 0.01 0

i

0

1O0

200 300 400 Temperature (K)

500

600

Fig. 8. Inverse magnetic correlation length for antiferromagnetic spin fluctuations in lightly doped La2 xSrxCuO4. The lines are based on a simple model in which the zero-temperature limit of t¢ is governed by the doping, and the temperature dependence is derived from the renormalized classical expression (eq. 5). From Keimer et al. (1992).

the decrease that occurs on cooling saturates at a doping-dependent value (see fig. 8). The increase in t¢ with temperature follows the same renormalized classical form (eq. 5) as the undoped material, leading to a simple description where the antiferromagnetic correlations in La2 xSrxCuO4 are characterised by an inverse correlation length that varies as to(x, T ) = it(x,0)+ tc(0,T), with to(x,0) varying as (a/x/-x) 1 and tc(0,T) following the Hasenfratz-Niedermayer expression. The form o f the zero-temperature inverse correlation length implies that the growth o f correlations is limited by the mean spacing between impurities. The shorter characteristic length scale associated with lightly doped La2CuO4 is also seen in the magnetic dynamics. As the energy transfer is increased, scans at constant energy remain centered on the (z~,Jr) position with a width that is always broader than the instrumental resolution (Hayden et al. 1991a). The width, an inverse length scale characteristic o f times ~l/co, increases linearly with energy transfer. This occurs much more rapidly than would be seen if the velocity o f propagation were the same as spin waves in LazCuO4: there is a softening to approximately 60% o f this value for the Lal.95Ba0.05CuO4 sample studied by Hayden et al. (1991a). The momentum dependence is not significantly modified by raising the temperature to above the spin-glass transition and beyond. The energy dependence o f the local (Q-integrated) susceptibility is sensitive to temperature at low energies. Figure 9 shows some data extracted from constant-Q scans (which are a good measure o f the Q-integrated cross-section over the limited range of ~o probed) which are well described by the form 1

Above the cross-over from localized to conducting transport (T ~ 50K), the inverse lifetime F satisfies F ~ T. This implies that the spin dynamics exhibit w / T scaling, a result

294

T.E. MASON i

La195 Baoo5Cu 04

Ef =8.09 meV 90'-60'- 52'-66'

c

7_

~oo

(a)

T=36 K

0

50

k,

• ~=11,o52,o)

"~

o Background

i =

0

(b) 2C -T=IOK 0

_

~

~

-

_j

I

I

-20

'{+

(c) ,+, ~t,÷,++÷÷~t,----, ~, ,,.,.o--~,

" T=64K

-20 20: 0

T=164K . ~ ~ . , ~ ~p.4.--r-

(d) ~4'**

•

-20 -5

"he (meVl

Fig. 9. Constant-Q scans at various temperaWres for Panel (a) shows the intensity at a 2D antiferromagnetic wave vector and displaced away in reciprocal space, to provide a measure of the background. Panels (b)-(d) show the measured energy-dependent response obtained by correcting the raw data for the background and thermal factor. The lines are fits to the single-site response discussed in the text. From Hayden et al. (1991a). Lal.95Sr0.05CuO 4.

also found by Keimer et al. (1991, 1992) where the inclusion o f data at low temperatures, below which logarithmic resistivity is seen, necessitates the inclusion of a higher-order, oo/T 3, term. This sort o f scaling form for the imaginary generalised susceptibility is consistent with the marginal Fermi liquid hypothesis which has been used to account for many of the anomalous normal-state properties o f the cuprates (Varma et al. 1989). The overall form taken by the temperature evolution o f X"(Q, oJ) indicates that the relaxation rate probed by Cu N M R is driven by the inverse lifetime, F, not by the correlation length, ~, as derived from scaling arguments (Millis et al. 1990). 2.3. Antiferromagnetism in YBaeCu306 +x The YBa2Cu306+x system is the most extensively studied o f all o f the families o f hightemperature superconductors, and has also been the topic of ongoing examination by inelastic neutron scattering. In this material, the antiferromagnetically ordered parent compound is YBa2Cu306. Superconductivity occurs when holes are introduced into the

NEUTRON

SCATTERING

STUDIES OF SPIN FLUCTUATIONS

IN HTSC

295

YBazCua06+ x 600

' I ' ' ' I'

' ' I ' ' ' ]'

' ' I'

' ' t-

500 400

f

!,

_

~

•

TN

o Tc 300

©

#

200

0

I F

100 , i,

0.0

#6

t,,

0.2

,~-1#?,

0.4

#7

,t 7,v

0.6

0.8

x

Iq

1.0

F i g . 10. P h a s e d i a g r a m o f Y B a 2 C H 3 0 6 + x s h o w i n g t h e d o p i n g d e p e n d e n c e o f t h e ant i f e r r o m a g n e t i c a n d s u p e r c o n d u c t i n g transitions. F r o m T r a n q u a d a et al. (1989).

CuO2 planes by adding additional oxygen. The rough phase diagram in fig. 10 shows the suppression of TN and the onset of superconductivity which occurs for x ~ 0.45. The magnetic dynamics of the Cu 2÷ moments in insulating YBa2Cu307_x are governed by the same Hamiltonian as La2CuO4(eq. 2). The principal difference has its origin in one of the distinguishing structural features of YBa2Cu306+x, the CuO2 bilayers. The close spacing of the CuO2 planes in a bilayer (3.2 A) gives rise to an inter-layer, intra-bilayer coupling U ± ~ 0). The inter-bilayer coupling is much smaller and can be neglected. The presence of the second term in the Hamiltonian leads to two branches in the spinwave dispersion which differs according to whether spins in adjacent layers rotate in the same ("acoustic") or different ("optic") directions (Tranquada et al. 1989). The dynamical susceptibility for the acoustic mode is

Za:(Q, to)= Zxxg2 lA2S ( 1 - Y(Q+)I y(Q) + J±/JI[) ~/2sin2 ( ~ ) g 3 ( h ° ) : k h ° ) a c ( Q ) ) , (7) and that for the optic mode is

Z~o~p(Q,~o) = ZzJvgZt~S

1+ ~/~±/J[I

cos 2 ~ - -

~ (he9 ± hcOop(Q)).

(8) The dispersion relations for the two modes are

hco (e) = 2Ji] {1 - y2(Q)+ J±/Jll [1 ± y(Q)]}l/2,

(9)

ac

op

where y(Q) = 1 [cos(23zh) + cos(2Jrk)] and the separation of the CuO2 planes in a bilayer, Az= 3.2 A (following conventional usage reciprocal space for YBa2Cu306+x is labelled

296

T.E. MASON

'

'

YBa2CU306.15

(T=295 K)

'

I

'

'

'

I

'

'

'

I

'

'

I l

,

,

,

I I

I

,

,

I I

,

,

(a) acoustic

,.-,, 81 "-~'aj '

,

'

I

,

l I

/

'/

, /,

j

(b) o p t i c a l [ @ ~

a

0 1 ,

o

i

C~1~31 , , , , , , , , , , , 50 100 150 200 hco ( me V )

Fig. 11. Energy dependence of the local susceptibility ofYBa2Cu306is for the (a) acoustic and (b) optic spinwave modes. The optic mode gap of 74 meV implies an inter-planar coupling of-11 meV From Hayden et al. (1998).

according to the tetragonal notation in which the antiferromagnetic ordering occurs at wavevectors related to (g,g,l), ~1 equivalent to (~,~) in the notation of the 2D square lattice reciprocal space). The inter-planar coupling does not introduce any dispersion along the c-axis; however, the sin 2 and cos 2 modulations can be used to distinguish between the two modes (for a more detailed treatment of the spin-wave theory, including additional, smaller, terms in the Hamiltonian, see Tranquada et al. 1989). The optic mode gap at the magnetic zone center has a magnitude given by hCOg=2 J V / J ~ , the inter-layer coupling is weak, but the intra-layer superexchange is comparable to the value found for La2euO4. Due to the relatively large value of JII, initial studies of YBa2Cu3 06 +x for antiferromagnetic compositions were not able to probe high enough frequencies to resolve the acoustic spin wave or observe the optic mode gap. Estimates of the in-plane antiferromagnetic exchange extracted from resolution-corrected fits to unresolved spin waves ranged from 80 meV to 200meV (Rossat-Mignod et al. 1988, 1991, Tranquada et al. 1989, Shamoto et al. 1993). The high-energy spin waves in YBa2Cu306.1s have now been resolved using the HET spectrometer (Hayden et al. 1996b, 1998) and yield a value of the exchange constant, J~ = 125-4-5 meV, obtained from measurements of the spin waves throughout the Brillouin zone. This is consistent with the value extracted from the interpretation of two-magnon Raman scattering (Singh et al. 1989). The same experiment also determined the quantum renormalization of the spinwave amplitude and found Z x = 0.44-0.1, in agreement with the La2CuO4 data (Hayden et al. 1996a) and the 1/S expansion (Igarashi 1992a,b). The optic mode, corresponding to out-of-phase motion of spins in adjacent layers of a bilayer, has also been observed by both reactor-based (hot source) triple-axis (Reznik et al. 1996) and spallation source timeof-flight (Hayden et al. 1996b) techniques. Figure 11 shows the energy dependence of the local susceptibility for values of c-axis momentum transfer chosen to probe the acoustic and optic mode response. The observed gap energy of 74±5 meV implies bilayer coupling

NEUTRON SCATTERINGSTUDIESOF SPINFLUCTUATIONSIN HTSC

297

close to band-theory calculations (Anderson et al. 1995) but about twice as large as the values obtained for Y2Ba4Cu7OI5 (Stern et al. 1995) and Bi2Sr2Ca2Cu3Ol0 (Statt et al. 1997) using spin-echo double resonance (a nuclear magnetic resonance technique). As was the case for La2CuO4, renormalized spin-wave theory gives a good description of the magnetic dynamics of antiferromagnetic YBa2Cu3 06 +x. The magnetic Hamiltonian for both compounds is the same (eq. 2) 2D spin-½ Heisenberg antiferromagnet. However, the bilayer nature of the YBa2Cu306+~ structure introduces an additional energy scale into the problem that is manifested in the optic spin-wave modes. The exchange energy between nearest-neighbour spins within a CuO2 plane is somewhat smaller than that found in La2CuO4, possibly reflecting the larger lattice constant. The coupling between layers in a bilayer is roughly 10% of the in-plane exchange.

3. Superconducting La2 xSrxCu04 Doping the single-layer system La2-xSrxCuO4 to compositions where superconductivity occurs produces dramatic changes in the energy, momentum, and temperature dependence of the magnetic excitations. Although it does not possess a particularly high Tc at optimal doping (~39 K), it has been extensively studied because of the relative simplicity of its crystal structure, without bilayers and chains, and the availability of reasonable large single crystals (grown by the travelling solvent zone method) which exhibit bulk superconducting transitions with To values close to those observed in the best powder samples. 3.1. Normal-state spin fluctuations The normal states of the metallic cuprates are, in many respects, as anomalous as their superconducting state. The spin fluctuations at low energies in La2 xSrxCuO4 undergo a qualitative change once the doping is sufficiently large to induce metallic behaviour at low temperatures. The mean impurity-spacing-limited, finite correlation length, antiferromagnetic response described in sect. 2.2 is replaced by an incommensurate magnetic structure which, for short time scales, has a much longer characteristic length. 3.1.1. Low-energy incommensurate response The low-energy spin dynamics of La2_xSrxCuOa tmdergo a qualitative change as the doping level exceeds that required to produce metallic conductivity (and, ultimately, superconductivity). The broad, commensurate response of the lightly doped compositions (see sect. 2.2) is replaced by an incommensurate response peaked at wavevectors displaced from the antiferromagnetic wavevector, (~,:v). The first indications of this were observed for samples with x=0.11 and x=0.15 (Birgeneau et al. 1989, Shirane et al. 1989, Thurston et al. 1989). The limitations of sample size and quality prevented determination of the location of the peak response in the planar reciprocal lattice, the temperature dependence and effect of superconductivity, and the lack of interplanar correlations.

298

T.E. MASON (a) 300 C

F:

250

I

I

I

La 1.925Sro.075CuO4 ]-I,

3 - Axis rico = 1 m e V

L

>P- 2 0 0 z LU Z

150

(b) 8£ "E" '~_ r--

60 -

(D

La1.86Sr0.140uO4 2 - Axis E i = 18.5 m e V T= 12K

-~ 40 >I-z 20 u.I I--z --

0

-

20

I

0.6

I

0.8

1.0

1.2

1.4

q x ( ~ ' lr)

Fig. 12. Low-energy incommensurate magnetic response in metallic La2 ~Sr,CuO4 for two compositions (x=0.075 and x=0.14). The data are scans in momentum transfer through two of the four peaks that appear at (~,~)zE6(~,0) and (~,~)±6(0,Jv). The upper panel is for an energy transfer of 1meV, while the lower is energy-integrated using an incident neutron energy of 18.55 meV From Cheong ct al. (1991).

Later work by Cheong et al. (1991) showed that the magnetic response was peaked at wavevectors (sr,~)i6(~,O) and (~,~)-4-6(0,~) in the notation o f the 2D square reciprocal lattice (see fig. 12). The magnitude o f the incommensuration, 6, scales approximately as 2 × and the intensity is rapidly suppressed as the temperature was raised to 100 K. In addition, superconductivity resulted in dramatic changes in the low-energy excitations described in sect. 3.2. The peaks are quite sharp in momentum space, indicating that on time scales comparable to the inverse o f the frequencies probed ( 1 - 1 5 m e V ) the correlations exist over distances larger than the mean spacing between impurities which limits the static correlations in lightly doped La2 xSrxCuO4. The doping dependence o f the incommensuration has been studied over a wider range o f compositions by Petit et al. (1997) and Yamada et al. (1998). The dependence, shown in fig. 13, is similar to that o f Tc (see fig. 2), although there are samples with no superconducting transition

NEUTRON SCATTERINGSTUDIES OF SPIN FLUCTUATIONSIN HTSC

0.15

[]

÷

299

+

,.--, 0.10 O

St-doping

~>

Oxygen-doping

[]

0.05 j'

Lal.6Ndo.4SrxeuO4

I~

Zn-substitution

O

Oxygen-reduction

,, i

0

,

0

,

,

,

I

,

i

i

,

I

,

,

,

J

I

,

0.1

,

,

,

I

0.2

,

,

,

,

I

i

i

i

i

0.:

Xeff Fig. 13. Incommensurability, 6, of the spin fluctuations in La2 .~Sr~CuO4 and related systems, plotted as a function of the effective doping level. The magnitude of the incommensuration scales in roughly the same way as the superconductingtransition temperature with doping. From Yamada et al. (1998). Note that the notation used to describe the location of the peaks in this figure (and the original reference) differs from that used in the text by a factor of 2.

that have an incommensurate response, indicating that it is not uniquely present in superconductors. Note that the notation used in fig. 13 and Yamada et al. (1998) is different from that o f Petit et al. (1997) and Cheong et al. (1991); hence, the differing scale o f the incommensuration, 6 ~x: the data are all consistent. The most notable feature of the behaviour seen in fig. 13 is the abrupt change from commensurate to incommensurate fluctuations for x ~ 0.06, making it coincident with the metal-insulator transition. There is also a saturation o f 6 beyond optimal doping (x = 0.14-0.17). Full exploration of higher doping has been precluded by the increasing difficulty o f growing homogeneous single crystals with x > 0.2. The origin o f incommensurate spin fluctuations occurring as a result o f the introduction o f holes into the CuO2 planes of La2CuO4 can be explained by a variety of models. The fact that the incommensurate response develops abruptly as the metallic compositions are encountered suggests that it might be a manifestation o f the Fermi surface. Incommensurate peaks in the dynamical susceptibility occur quite naturally for a 2D square lattice metal away from half filling and can be calculated on the basis of a tightbinding model of the band structure (Bulut et al. 1990, B6nard et al. 1993, Littlewood et al. 1993, Si et al. 1993, Tanamoto et al. 1994). This description has the virtue that it leads to a simple interpretation o f the effects o f superconductivity on the spin fluctuations. However, there is not yet detailed agreement between the experiments and

300

T.E. MASON

Q = Qx (re, ~) + (8/2) (~, if,) T=35K RIS(~ TAS6

o ,q,

150

"~ 100 >-

• ,

0.6

0.8

'

1.0' ' '1.2 Qx

. . . . .

(2n, o)

1.4 Qz

(0, 0)

Fig. 14. Energy and momentum dependence of the incommensurate spin fluctuations in the normal state of Lal.s6gr0J4CnO4. The inset in the lower left shows the scan trajectory in the 2D square lattice momentum space (dashed line). The low-fi'equency incommensurate response broadens with increasing energy, merging into a broad commensurate feature as the energy is increased beyond ~15 meV From Mason et al. (1992).

calculations in either the normal or superconducting state. An alternative view is that the incommensurate peaks are a consequence of commensurate antiferromagnetic correlations within stripes delineated by the ordering of the charged holes (Emery and Kivelson 1993, Zaanen and Gunnarsson 1989). Static charge and spin ordering of this type occurs in doped La2NiO4 and Nd-doped La2_xSrxCuO4 (for a recent review, see Tranquada 1998). For superconducting La2_xSrxCuO4 there is no static order, and attempts to observe charge peaks related to the incommensurate magnetic peaks have not been successful to date. However, the similar wavevector of the magnetic modulations in the ordered systems to the incommensurate peak positions in the superconducting compounds is suggestive. Descriptions taking the antiferromagnetic parent compound as a starting point and introducing frustration with the holes can also lead to incommensurate instabilities (Aharony et al. 1988, Shraiman and Siggia 1989). The evolution of the incommensurate response as the energy is increased is summarised in fig. 14 which shows scans through two of the four incommensurate peaks in La].86Sr014CUO4 at 3.5, 6, 9, 12, and 15 meV at 35 K (just above To). The incommensurate peaks are well defined at lower energies but broaden with increasing energy; the scan at 15 meV no longer displays well-resolved incommensurate features. Qualitatively, this (E,Q) dependence is very similar to that observed in Cr and Cr alloys above the spin density wave (SDW) transition (Noakes et al. 1990). The lines in fig. 14 are the results of a fit to a functional form proposed by Sato and Maki (1974) to describe Cr (modified in

NEUTRON SCATTERINGSTUDIESOF SPIN FLUCTUATIONSIN HTSC

301

this case to take account of the two-dimensional nature of the incommensurate magnetism in La2 xSrxCuO4). More recent results (Aeppli et al. 1997b) with better signal/noise and improved counting statistics have shown that this simple, mean-field, model is not sufficient to simultaneously describe the momentum and energy dependence of the spin fluctuation in the normal state of Lal.86Sr0.~4CuO4. However, the general trend of the low-energy peaks to merge into a broad, flat-topped, commensurate response at higher energies (which also occurs in the Cr systems) is captured. Beyond ~25 meV, the incommensurate character of the magnetic response disappears (Petit et al. 1997), although the very interesting regime where this incommensurate-commensurate collapse occurs has not been extensively explored due to the presence of optical phonons, making unpolarized measurements unreliable. The high-energy response, which in many respects resembles that of La2CuO4, will be discussed in sect. 4.1.2. For energies below 15meV, the spin dynamics are well described by a generalisation of the Sato and Maki (1974) form:

[~(~0) + 1]. x~;'(~o,r ) . ~'4(c0, r) S(Q, w) =

[I¢2(~, T) + R(Q)I 2

,

(10)

where R(Q) = [(qx-qy)2-(~6)2]

2+ [(qx+qy)2-(Jr6)2] 2

2 (2a0;r6) 2

(11)

is a function, with the full symmetry of the reciprocal lattice and dimensions of IQ[ 2, which is positive except at zeroes, coinciding with the incommensurate peak positions. The parameters t¢((o, T) and Xi~((o,T) describe the frequency- and temperature-dependent inverse length scale and peak susceptibility, respectively. This form was chosen because it provided a good description of the data (when corrected for resolution effects) over a wide range of temperature (35-350K) and energy (1 15meV). In particular, the strong 1/q 4 drop-off in intensity away from the incommensurate peaks is required to describe the lower-temperature data (a Lorentzian does not work). In addition, the shape of R(Q) reproduces the tendency of the data to exhibit extra intensity between the incommensurate peaks, particularly as the energy (or temperature) is increased. The results of this analysis are summarised in fig. 15 which shows the evolution of the inverse length scale t¢((o, T) plotted against temperature and energy added in quadrature for energies between 2.5 and 15 meV and temperatures between 35 and 350 K. All of the data follow the simple form

t¢2 = t¢2 + ao 2 [(kBT/Er)2 + (hoJ/E(o) 211/z ] ,

(12)

where Z = 1, t¢o=0.034 ~ , and ET-E~o = 47 meV (~J/3), shown as a solid line. The inset in the upper right of fig. 15 shows the variation of the low-frequency limit of X~/o with

302

T.E. MASON

104 103 102 v

0,20 ~z

0,15 0.10 0.05 0

0

1O0 200 ~(T2+(h co/kJ ) ( K )

300

400

Fig. 15. Temperature dependence of the inverse length scale, I¢(e~,T) at various fixed energy against temperature and energy, added in quadrature. The solid line corresponds to a Z = 1 description of the data. The graph in the upper right shows the t¢ 3 variation of the peak response. The schematic phase diagram in the upper left depicts a 3D phase space with a

transfers plotted quantum-critical incommensurate quantum-critical

point at a. The experimentstake place in the dark vertical plane slightly displaced in phase space from the critical point. From Aeppli et al. (1997b). ~c which obeys a power law t¢-a with &= 3. The (co, T) scaling, dynamical exponent Z = 1, and 6 = (2-~/+Z)/Z = 3, are all consistent with a system close to a magnetic instability the Quantum Critical Point (QCP) of a 2D insulating magnet (Sachdev and Ye 1992, Chubukov et al. 1994, Millis 1993). The eventual saturation of t¢ at a finite value t¢o in the zero T and co limit implies an offset from the QCP in phase space, perhaps along the lines of the schematic representation in the upper left of the figure. The similarity of the wavevector of the dynamic incommensurate structure observed in La2_xSrxCuO4 to the static charge and spin ordering in compounds nearby in parameter space suggests stripe formation may be the driving mechanism. It has been proposed that this is an explanation for both the anomalous normal-state properties of the cuprate superconductors and the underlying origin of superconductivity (Castellani et al. 1998). 3.1.2. Comparison of high-energy spin dynamics with La2Cu04 The incommensurate fluctuations that typify the magnetic response of the metallic and superconducting compositions of La2 xSrxCuO4 are a low-energy feature, merging into a broad commensurate response above ~20 meV, similar to that seen in lightly doped insulating samples. Since the magnetic response of insulating La2 xSrxCuO4 extends over energies up to 320meV one might expect a similar energy scale to be relevant in the superconducting compositions. This issue has been explored using the HET spectrometer at ISIS by Hayden et al. (1996a, 1998). Figure 16 shows a comparison of the energy and frequency dependence of the response in La2CuO4 and Lal.86Sr0.14CuO4 with the intensity

NEUTRON SCATTERINGSTUDIES OF SPIN FLUCTUATIONSIN HTSC

30 20 10 0

v'-"

] La2Cu04 ]

I La1.86Sro.14Cu04 I

(a)

(g)

~

0

~ x 2.5

100-150 meV

x 2.5

I

30 20 10 0

I

2 30

~

meV

303

I

[

I

L

2o

,~m 10 :::L 0 o 30 co 20 10 0 30 20 10 0

_

50-75meV

- ~&~ 0.5

1

~ 1.5 0.5 h (r.l.u.)

] " 1

1.5

Fig. 16. Magnetic scattering from L a 2 C u O 4 (~f) and from Lal.a6Sr0.14CuO4 (~1) between 25 and 200meV. All scans are in the same absolute units. The overall energy scale in the two systems is comparable, but the magnetic peak is broader in the superconductor and the intensity is suppressed at higher frequencies. From Hayden et al. (1998).

units normalised to allow a detailed comparison. The schematic diagrams at the top o f the figure give an impression o f how the spectrum is modified by doping. The response emanating from the commensurate position is broader (reflecting shorter characteristic length scale) in momentum and falls off more rapidly with increasing frequency. The difference in the two spectra is most clear in the local susceptibility, obtained by integrating the response in momentum space. Figure 17 shows this comparison for the same two samples. The doped compound has spectral weight over the same large range of energies as the antiferromagnetic insulator, but there is a significant redistribution of spectral weight from high energy to a peak in the response at ~ 2 0 - 3 0 meV. In contrast, for the antiferromagnet the local susceptibility is peaked at the zone-boundary energy (not shown) due to the van Hove singularity in the spin-wave density of states. Interestingly,

304

T.E. MASON [

T

1

(a) La2CuO 4 (T = 2 9 5 K )

3 2 'T> cdrn

1 0 4

3 2 1

0

0

50

100

150

he) (meV)

200

250

Fig. 17. Local (Q-integrated) susceptibility for antiferromagnetic and superconducting La2 xSrxCuO4. The energies over which magnetic correlations exist are comparable for x=0 and x 0.14. However, the doped sample displays a low-energy peak in the specmam at -20-30meV which is not present in the insulator (where the spectrum is maximum at the zone-boundary energy due to the van Hove singularity in the density of states). From Hayden et al. (1998).

the peak in the response in the x = 0 . 1 4 sample occurs at the same energy that the incommensurate peaks merge into a broad commensurate feature. 3.2. Superconductivity and the incommensurate spin fluctuations If the low-energy incommensurate peaks in La2 x S r x C u O 4 c a n be thought of as arising from transitions across the Fermi surface in which a quasiparticle moves from an occupied to an unoccupied state, at the same time flipping its spin, then the superconducting transition should have a dramatic effect since such a particle-hole excitation will acquire a gap o f 2A. In light o f this it is perhaps not surprising that the low-energy spin fluctuations are suppressed by the onset o f superconductivity. However, this simple physical picture ignores much of the complexity o f the phenomena that were first observed by Mason et al. (1992). This effect has now been studied in detail for a number of superconducting samples. The effect o f superconductivity on the incommensurate response is summarised in fig. 18, which shows both the low-energy suppression and the associated enhancement at somewhat higher frequencies. The cross-over between suppression and enhancement may be thought o f as a measure o f the superconducting gap (actually twice the gap) at the wavevectors connected by the Q probed in the simple picture. For the incommensurate wavevector this energy is 7 meV Closer scrutiny o f the data reveals more subtle behaviour than might be expected on the basis o f the non-interacting Fermi surface particle-hole analogy. For a start, the suppression o f the low-energy excitations is not complete for the 6.1meV data shown in fig. 18, (the same is true down to 1.5meV) - in a clean conventional superconductor there should be no particle-hole excitations below 2A. The size of the residual low-energy response is sample dependent, ranging from nearly 100% o f the normal state response in samples with deliberately introduced Zn impurities (Matsuda et al. 1992) to nearly total suppression for higher-quality samples (Yamada et al. 1995), suggesting an impurity-induced low-energy response in the superconducting

NEUTRON SCATTERING STUDIES OF SPIN FLUCTUATIONS IN HTSC

305

Tc i

+ , + + +

60

_ '

15meV

5O %0~ -~ 40

I

E

I

c

E

¢,D

9 meV

65 s5

rn

45

I

Q=(n,n)+E)(n,O) /

E 80

o I/(n(e))+l)

O

o -~ 70

,

I

6.1 meV

/,,, \\\

60 50 40

i

0

40

20

T(K)

60

Fig. 18. Temperature dependence of the incommensurate magnetic response in Lal 86Sro.14CuO 4 for 6.l, 9, and 15meV energy transfer. For energies below 7 meV, the spin fluctuations are suppressed, while above that energy, they are enhanced, in the superconducting state. From Mason et al. (1996).

state. O f course, a response at low energies does occur in a d-wave superconductor and therefore might well be expected in a high-To material. However, the calculations based on a d-wave order parameter predict a modification o f the wavevector dependence o f the low-energy response (Zha et al. 1993) which has not been observed (Mason et al. 1993). Another feature o f the data which has not been accounted for theoretically is the fact that the enhancement below Tc for energies above 7 m e V is restricted to a very narrow region o f Q near the incommensurate peak, essentially limited by spectrometer resolution. This implies a very long coherence length, in excess o f 50 A, which is greater than any length scale present in the normal state.

4. S u p e r c o n d u c t i n g Y B a 2 C u 3 0 7 _ x The initial measurements o f the low-energy spin fluctuations in YBa2Cu307 x for samples with oxygen content beyond the x~0.5 required for superconductivity revealed a peak, centered on the commensurate ( ~ , ~ ) position, which is considerably broader in momentum than in insulating compositions (Tranquada et al. 1990, Chou et al. 1991, Rossat-Mignod et al. 1988, 1991). The evolution o f the normal-state commensurate response as the doping is increased beyond x = 0.5 is summarized in fig. 19: the general trend o f a shift in the characteristic energy o f the scattering and overall decrease in

T.E. MASON

306

NormalState, 100 K, i

•

.

.

+

.

.

.

.

,

.

.

.

.

,

-

•

•

200

.

T 0

~0%i -: -

150

i,~

\

2

0

3 65 100

-

r"

150 "

.,.-; I,~;,

>, L ¢

YBa Cu 0 = 3 6`.

:

IO0 ;-

.Q

SO.

<

:0'-,-.., j

°,'°-

| J

/-./

i

150

0

N

Z"

T

'

'

YBa Cu 0

~

2

3 6.92 100

E

m

"i

C T

T

"+'9 '+ "

~"

i,

100

YBa Cu 0 Z

50

o r .T

-

-

0

10

20

30

Energy(meV)

3

6`97

[

.

°O Tg.e o,."T

0

T

t I

40

50

i

5O

Fig. 19. Low-energy, commensurate [Q = (zc,zc), acoustic ]node] response in the normal state of four different compositions of YBa2Cn307_, measured at 100K. As the doping is increased the feature at (~c,zc) broadens and weakens, and there is very little normal state response at the commensurate position for the overdoped sample. From Bourges (1998).

intensity as the p e a k smears out in Q - s p a c e m a k e s the m e a s u r e m e n t s increasingly difficult - to the point that for o v e r d o p e d samples the (:v,~) response is barely observable. The c-axis m o d u l a t i o n due to the bilayer c o u p l i n g w h i c h leads to an optic and acoustic m o d e in the spin-wave cross-section for antiferromagnetically ordered samples is also

NEUTRON SCATTERINGSTUDIES OF SP1N FLUCTUATIONSIN HTSC YBa2Cu306.

6 (T=80

201 .... i .... i .... 15 / (a) a c o u s t i c

i ....

K) i ....

~m o i ~ " ' ; - ' ' ; - ' ¼ " H ' ; =- _ | (b) optic

....

,oL A, 0

.

0

50

.

100

307

.

.

150

h f9 ( m e V )

.

200

250

Fig. 20. Normal-state local (Q-integrated) response for YBa2Cu306.6 as a function of energy transfer. The energy gap associated with the optical mode is comparable to that observed for the spin waves in antiferromagnetic, insulating YBa2Cu306.15.From Hayden et al. (1998).

present for the fluctuations in metallic samples. As was the case for the spin waves discussed in sect. 2.3, the low-energy response is the acoustic mode, peaked at the c-axis wavevector that corresponds to a Jr phase shift within a bilayer. By selecting the c-axis momentum transfer, the acoustic and optic response can be probed separately, giving a determination of the optic gap, 43 meV, comparable to the spin-wave gap for the optic mode (Hayden et al. 1998, Reznik et al. 1996). Figure 20 summarizes this result for x = 0.4, showing the local (Q-integrated) susceptibility up to 200 meV (similar results have been obtained for x =0.5 (Bourges 1998); including the unexplained second peak in the acoustic response at the same energy as the peak in the optic spectrum). The most striking feature of the data shown in fig. 20 is the significant difference in overall energy variation and the absolute intensity compared to that of fig. 11 for the spin waves in a sample with x = 0.85. There is a shift in spectral weight to lower frequencies in the metallic sample.

4.1. Low-energy incommensurate response The fact that initial measurements o f the inelastic magnetic response o f superconducting compositions o f YBa2Cu;O7 x found only a broad commensurate peak at low energies (in contrast to the incommensurate response found in La2_xSrxCuO4) was interpreted as arising from differences in the Fermi surface geometry o f the two compounds (Si et al. 1993). There were indications that the Q-dependence of the cross-section for underdoped YBazCu307-x was better described in terms of unresolved incommensurate peaks, manifested as a flat-topped peak at (~,zc) which was not well described by a single Gaussian (Sternlieb et al. 1994). More recently Dai et al. (1999) have shown for a sample with x = 0.4 (the same one as used for the data shown in fig. 20) a clear incommensurate response. Improvements to the HET spectrometer at ISIS, in the form o f an array o f position-sensitive detectors at low angles, have allowed a complete mapping o f the incommensurate structure (see fig. 21). This feature is only visible at low temperatures ( T = 13 K and 65 K for the data shown) and energies below - 3 0 meV, as was the case for

308

T.E. M A S O N Global Fit

YBCO 0.7 0.6 =, ,~ 0.5 0.4 a 0.3 0.3

0.4

0.5

0.6

0.7

0.3

T= 13 K, E=34 meV

0.4

0.5

0.6

0.7

T=13 K, E=24.5 meV

0.3

0.4

0.5

0.6

0.7

T=65 K, E=24.5 m e V

0.3

0.4

0.5 0.6 h (r.l.u.)

0.7

0,3

0.4

0.5 0.6 h (r.l.u.)

0.7

Fig. 21. Images of the magnetic scattering from YBa2Cu306. 6 above and below T c at 34 and 24.5meV in the two-dimensional reciprocal space of the CuO2 planes. At the lower energy (e,f) an incommensurate response, described by the model shown in (d), appears at the positions noted in the schematic map (a). The resonance that appears at the (~,Jr) position (b,c) in the superconducting state is described in sect. 4.2. From Mook et al. (1998a).

La2 xSrxCuO4.Its emergence is not connected with the superconducting transition but is simply a sharpening as the temperature is lowered. The locations of the peaks are the same as for equivalently hole-doped Lal.9Sr0.tCuO4 (Mook et al. 1998a). The discovery that the low-energy spin fluctuations in YBazCu3OT-x, at least for underdoped samples, have the same incommensurate structure as Laz_xSrxCuO4 suggests that this is a universal feature of high-Tc materials. It is also consistent with current understanding of the Fermi surface topology in both systems (see ch. 201). How the incommensuration evolves as the oxygen content is increased towards optimal-doped requires the type of detailed comparative study that has been carried out for La2 xSrxCuO4

NEUTRON SCATTERING STUDIES OF SPIN FLUCTUATIONS IN HTSC

309

(Yamada et al. 1998). The superconducting transition results in both a sharpening of the incommensurate peaks and an elimination of the lowest-energy fluctuations; identical behaviour is found in nearly optimally doped Lal.s6Sr0.14CuO4(Mason et al. 1996). 4.2. The (Jr,~) resonance Much of the interest in inelastic neutron scattering from high-temperature superconductors has been focussed on the dramatic effects below T~ often referred to as the resonance peak. Rossat-Mignod et al. (1991) first observed a sharp, intense peak at (3v,Jr) that appeared below T~ in optimally doped 123. The detailed energy dependence was complicated by the presence of phonons, nearby in energy and momentum. The magnetic origin of the 41 meV resonance was confirmed for x=0.07 by Mook et al. (1993) using polarized neutrons (see fig. 22). This allowed the magnetic and lattice contributions to be separated and verified that the dominant feature in the magnetic 50 ~ Z,(¢o) T=100;(

~

SPIN FLIP -

20

i 70""

60 ~-

Z'(o~) T=IOK (0.5, 0.5, -5.2) (e, x)

f~

SPIN FLIP

_

5O

40 z O :E 3 O o 20 8 z

0

p) 10o 80 6O 40 20

(c)

o

4

1oo

I

NON SPIN FLIP o 10K o 100K

a I

B^CKGR~e~I.o,0OK 10K .~,.,~,~ I

t

B tr

40 80

|(d)

I

I

6o 20

~ -20 --'3 40 8 -e0

(o) 0

I 2o

4o meV

6o

Fig. 22. Polarized-beam measurements of the magnetic excitations in YBa2Cu307 confirming the magnetic origin of the resonance feature that appears at the (Jr,~) position at 41 meV below T c. From Mook et al. (1993).

310

T.E. MASON

5O

r

E

40 30

13 e"

/

E

LU

20 7 r =0

10

13

/

~

I / i,,.J..

France

* Oakridge ~ 0 Princetoni

0 0

20

40

60

Tc(K)

80

100

Fig. 23. Variation of the (~,Jv) resonance energy with superconducting transition temperature, T c. From Bourges (1998).

response of the superconducting state was a sharp (in energy) peak at 41 meV, centered on the commensurate, (~,:r), position. As the oxygen content is decreased the energy of the resonance peak decreases, following the same trend as the superconducting transition temperature. For x = 0.4 the resonance energy has been reduced to 34 meV; the overall variation with T~ is summarized in fig. 23. There is a clear anomaly in the temperature dependence at the superconducting transition, with an abrupt jump in intensity above the level of a weaker normal-state response. The existence of a weak normal-state feature was disputed (see Fong et al. 1995); however, the different measurements were not performed on samples with the same oxygen content. The general trend of the normal-state response shown in fig. 19 indicates a gradual elimination of the intensity at (~,:c) in the normal state as the oxygen content is increased beyond optimal doping (x 0.07). For underdoped samples, where the normal-state intensity is quite significant, the presence of a peak at essentially the same energy as the resonance above Tc is clear (see fig. 24). The growth in the intensity at the resonance energy begins well above Tc although there is a rapid increase, and an associated reduction in energy width, which occurs just below To. The size of the peak in the normal state is reduced as the oxygen content is increased, and the temperature at which it becomes apparent is also reduced (this onset temperature appears to track the pseudo gap feature seen by NMR) (Dai et al. t999). There are a variety of theoretical descriptions of the origin of the (:-c,s~) resonance, ranging from a superconducting coherence effect which occurs in the framework of an itinerant-electron picture (Ohashi and Shiba 1993, Monthoux and Scalapino 1994, Lu 1992, Bulut and Scalapino 1996, Lavagna and Stemmann 1994, Onufrieva a~d Rossat-Mignod 1995, Liu et al. 1995) to a collective mode associated with a muiticomponent, SO(5), order parameter (Zhang 1997). In almost all descriptions of the resonance peak it is a direct consequence of the d-wave symmetry of the order parameter.

NEUTRON SCATTERING STUDIES OF SPIN FLUCTUATIONS 1N HTSC

311

YBC06.6 1200

500

F •

1000

~ OK

I

150K

400

800 600 400 200 0 c 0

E

300 -200 I00 0 500

600

200 K

0 0 400O'b

400 300 200

~m e'- 200 0 ~

0

°°

20 25 30 35 40 45 SO

400

~00

I

~00

300

300

200 I00

~

--

ao

_~oo I00

• as

30

3s

40

E(meV)

4s

J

so

3

zo

zs

30

3s

40

4s

so

E(meV)

Fig. 24. Temperature dependence of the 35 meV resonance in YBa2Cu306. 6 with temperature. A broadened response at (s%~c) persists in the normal state for underdoped compositions. From Mook et al. (1998b).

In many cases, however, the existence of a resonance above the superconducting transition poses a challenge.

5. Conclusions

The experimental difficulties posed by the large energy scale, weak spin-½ crosssection, and (for metallic samples) short characteristic length scale in high-temperature superconductors has necessitated advances in the state of the art in inelastic neutron scattering in order to develop an overall picture of the important features. The improved capabilities of neutron instrumentation, together with significant advances in crystal growth, particularly of large, high-quality crystals needed for this type of work, have led to a fairly consistent picture of spin fluctuations in the cuprates. Generally, when measurements on the same compounds (at the same doping level) exist, there is good agreement between different groups. The importance of these measurements to the field of high-temperature superconductivity as a whole relies on the special characteristics of

312

T.E. MASON

the neutron as a probe. It is the only technique which permits measurements o f spin dynamics at finite momentum, with good energy and Q resolution. The antiferromagnetically ordered insulators are well described by conventional spin-wave theory. The destruction o f long-range antiferromagnetic order with doping leads to novel scaling behaviour, in a spectrum which is still reminiscent o f the spin waves. Metallic cuprates have a low-energy response that is incommensurate and qualitatively quite different from the insulators. Furthermore, it is greatly affected by the superconducting transition. A t higher energies the changes are much less dramatic as a function o f doping and temperature. A l l o f these features are shared by both families that have been extensively studied, La2 xSixCuO4 and YBa2Cu307-x. The one outstanding difference is the (zc, Jr) resonance which has only been seen in YBa2Cu307 x. A search for a commensurate resonance in La2 xSr, CuO4 is probably warranted given the otherwise similar picture which has emerged o f the overall behaviour in comparison to YBa2Cu3 O7.

Acknowledgements I would like to thank my colleagues for allowing me to reproduce figures, providing material prior to publication, and stimulating discussions. I would particularly like to thank my collaborators in experiments I have been involved in for their patience and guidance. The financial support received from the Natural Sciences and Engineering Research Council o f Canada, the Canadian Institute for Advanced Research, and the Alfred P. Sloan Foundation is gratefully acknowledged. Work at Oak Ridge was supported by the US Department o f Energy under Contract DE-AC05-00OR22725.

References Aeppli, G., S.M. Hayden, H.A. Mook, Z. Fisk, S.W Cheong, D. Rytz, J.R Remeika, G.P. Espinosa and A.S. Cooper, 1989, Phys. Rev. Lett. 62, 2052. Aeppli, G., S.M. Hayden and T.G. Perring, 1997a, Physics World 33(December), 33. Aeppli, G., T.E. Mason, S.M. Hayden, H.A. Mook and J. Kulda, 1997b, Science 278, 1432. Aharony, A., R.J. Birgeneau, A. Coniglio, M.A. Kastner and H.E. Stanley, 1988, Phys. Rev. Lett. 60, 1330. Anderson, O.K., A.I. Liechtenstein, O. Jepsen and F. Paulsen, 1995, J. Phys. Chem. Solids 56, 1573. B6nard, R, L. Chen arid A.-M.S. Tremblay, 1993, Phys. Rev. B 47, 15217. Birgeneau, R.J., J. Als-Nielsen and G. Shirane, 1977, Phys. Rev. B 16, 280.

Birgeneau, R.J., Y. Endoh, K. Kakurai, Y. Hidaka, T. Murakami, M.A. Kastner, T.R. Thurston, G. Shirane and K. Yamada, 1989, Phys. Rev. B 39, 2868. Birgeneau, R.J., A. Aharony, N.R. Belk, EC. Chou, Y. Endoh, M. Greven, S. Hosoya, M.A. Kastner, C.H. Lee, Y.S. Lee, G. Shirane, S. Wakimoto, B.O. Wells and K. Yamada, 1995, J. Phys. Chem. Solids 56, 1913. Bourges, Ph., 1998, in: The Gap Symmetry and Fluctuations in High Temperature Superconductors, eds J. Bok, G. Dentscher and D. Pavuna (Plenum Press, Cambridge) pp. 349-371. Bulut, N., and D.J. Scalapino, 1996, Phys. Rev. B 53, 5149. Bulut, N., D. Hone, D.J. Scalapino and N.E. Bickers, 1990, Phys. Rev. Lett. 64, 2723.

NEUTRON SCATTERING STUDIES OF SPIN FLUCTUATIONS IN HTSC Castellani, C., C. Di Castro and M. Grilli, 1998, L Phys. Chem. Solids 59, 1694. Chakravarty, S., B.I. Halperin and D.R. Nelson, 1989, Phys. Rev. B 39, 2344. Cheong, S.-W., G. Aeppli, T.E. Mason, H.A. Mook, S.M. Hayden, EC. Canfield, Z. Fisk, K.N. Clansen and J.L. Martinez, 1991, Phys. Rev. Lett. 67, 1791. Chou, H., J.M. Tranquada, G. Shirane, T.E. Mason, W.J.L. Buyers, S. Shamoto and M. Sato, 1991, Phys. Rev. B 43, 8690. Chnbukov, A., S. Sachdev and J. Ye, 1994, Phys. Rev. B 49, 11919. Collins, M.E, 1989, Magnetic Critical Scattering (Oxford University Press, Oxford). Dai, E, H.A. Mook, S.M. Hayden, G. Aeppli, T.G. Perring, R.D. Hunt and E Dogan, 1999, Science 284, 1344. Emery, VJ., and S.A. Kivelson, 1993, Physica C 209, 597. Endoh, Y., K. Yamada, R.J. Birgenean, D.R. Gabbe, H.E Jenssen, M.A. Kastner, C.J. Peters, RJ. Picone, T.R. Thurston, J.M. Tranquada, G. Shirane, Y. Hidaka, M. Oda, Y. Enomoto, M. Suzuki and T. Murakami, 1988, Phys. Rev. B 37, 7443. Fong, H.E, B. Keimer, RW. Anderson, D. Reznik, E Dogan and I.A. Aksay, 1995, Phys. Rev. Lett. 75, 316. Grempel, D.R., 1988, Phys. Rev. Lett. 61, 1041. Greven, M., R.J. Birgeneau, Y. Endoh, M.A. Kastner, B. Keimer, M. Matsuda, G. Shirane and T.R. Thnrston, 1994, Phys. Rev. Lett. 72, 1096. Greven, M., R.J. Birgeneau, Y. Endoh, M.A. Kastner, M. Matsuda and G. Shirane, 1995, Z. Phys. B 96, 465. Harshman, D.R., G. Aeppli, G.E Espinosa, A.S. Cooper, J.E Remeika, E.J. Ansaldo, T.M. Riseman, D.L1. Williams, D.R. Noakes, B. Ellman and T.E Rosenbaum, 1988, Phys. Rev. B 38, 852. Hasenfratz, E, and E Niedermayer, 1991, Phys. Lett. 268, 231. Hayden, S.M., G. Aeppli, H.A. Mook, S.-W. Cheong and Z. Fisk, 1990, Phys. Rev. B 42, 10220. Hayden, S.M., G. Aeppli, H.A. Mook, D. Rytz, M.E Hundley and Z. Fisk, 1991a, Phys. Rev. Lett. 66, 821. Hayden, S.M., G. Aeppli, R. Osborn, A.D. Taylor, T.G. Perring, S.-W Cheung and Z. Fisk, 1991b, Phys. Rev. Lett. 67, 3622. Hayden, S.M., G. Aeppli, H.A. Mook, T.G. Perring, T.E. Mason, S.-W. Cheong and Z. Fisk, 1996a, Phys. Rev. Lett. 76, 1344.

313

Hayden, S.M., G. Aeppli, T.G. Perring, H.A. Mook and E Dogan, 1996b, Phys. Rev. B 54, R6905. Hayden, S.M., G. Aeppli, E Dai, H.A. Mook, T.G. Perring, S.-W. Cheong, Z. Fisk, E Dogan and T.E. Mason, 1998, Physica B 241-243, 765. Igarashi, J., 1992a, Phys. Rev. B 46, 10763. Igarashi, J., 1992b, J. Phys.: Cundens. Matter 4, 10265. Imai, T., C.E Slichter, K. Yoshimara, M. Katoh and K. Kosuge, 1993, Phys. Rev. Lett. 71, 1254. Ivanov, A.S., Ph. Bourges, D. Petitgrand and J. RossatMignod, 1995, Physica B 213 214, 60. Johnston, D.C., J.P. Stokes, D.R Goshnrn and J.T. Lewandowski, 1987, Phys. Rev. B 36, 4007. Kastner, M.A., R.J. Birgeneau, T.R. Thnrston, EJ. Picone, H.P. Jenssen, D.R. Gabbe, M. Sato, K. Fukuda, S. Shamoto, Y. Endoh, K. Yamada and G. Shirane, 1988, Phys. Rev. B 38, 6636. Keimer, B., R.J. Birgenean, A. Cassanho, Y. Endoh, R.W. Erwin, M.A. Kastner and G. Shirane, 1991, Phys. Rev. Lett. 67, 1930. Keimer, B., N.R. Belk, R.J. Birgeneau, A. Cassanho, C.Y. Chert, M. Greven, M.A. Kastner, A. Aharony, Y. Endoh, R.W. Erwin and G. Shirane, 1992, Phys. Rev. B 46, 14034. Lavagna, M., and G. Stemmann, 1994, Phys. Rev. B 49, 4235. Littlewood, RB., J. Zaanen, G. Aeppli and H. Munien, 1993, Phys. Rev. B 48, 487. Liu, D.Z., Y. Zha and K. Levin, 1995, Phys. Rev. Lett. 75, 4130. Lovesey, S.W., 1984, Theory of Neutron Scattering from Condensed Matter (Clarendon Press, Oxford). Lu, J.R, 1992, Phys. Rev. Lett. 68, 125. Mason, T.E., G. Aeppli and H.A. Mook, 1992, Phys. Rev. Lett. 68, 1414. Mason, T.E., G. Aeppli, S.M. Hayden, A.E Ramirez and H.A. Mook, 1993, Phys. Rev. Lett. 71,919. Mason, T.E., A. Schr6dm, G. Aeppli, H.A. Mook and S.M. Hayden, 1996, Phys. Rev. Lett. 77, 1604. Matsuda, M., Y. Endoh, K. Yamada, H. Kojima, I. Tanaka, R.J. Birgenean, M.A. Kasmer and G. Shirane, 1992, Phys. Rev. B 45, 12548. Matsuda, M., K. Kuroda, K. Yamada, Y. Endoh, H. Chou, G. Shirane, R.J. Birgeneau, M.A. Kasmer, I. Tanaka and H. Kojima, 1993, J. Appl. Phys. 73, 6333. Millis, A.J., 1993, Phys. Rev. B 48, 7183. Millis, AJ., H. Monien and D. Pines, 1990, Phys. Rev. B 42, 167.

314

T.E. MASON

Mitsuda, S., G. Shirane, S.K. Sinha, D.C. Johnston, M.S. Alvarez, D. Vaknin and D.E. Moncton, 1987, Phys. Rev. B 36, 822. Monthoux, P., and D.J. Scalapino, 1994, Phys. Rev. Lett. 72, 2131. Mook, H.A., M. Yethiraj, G. Aeppli, T.E. Mason and T. Armstrong, 1993, Phys. Rev. Lett. 70, 3490. Mook, H.A., P Dai, S.M. Hayden, G. Aeppli, T.G. Perring and E Dogan, 1998a, Nature 395, 580. Mook, H.A., R Dai, R.D. Hunt and E Dogan, 1998b, J. Phys. Chem. Solids 59, 2140. Noakes, D.R., T.M. Holden, E. Fawcett and EC. Decamargo, 1990, Phys. Rev. Lett. 65, 369. Ohashi, Y., and H. Shiba, 1993, J. Phys. Soc. Jpn. 62, 2783. Onufrieva, E, and J. Rossat-Mignod, 1995, Phys. Rev. B 52, 7572. Petit, S., A.H. Moudden, B. Hennion, A. Vietkin and A. Revcolevschi, 1997, Physica B 234-236, 800. Reznik, D., Ph. Bonrges, H.E Fong, L.P. Regnault, J. Bossy, C. VeRier, D.L. Milius, I.A. Aksay and B. Keimer, 1996, Phys. Rev. B 53, R14741. Rossat-Mignod, J., R Butler, M.J. Jurgens, C. Vettier, L.P. Regnault, J.Y. Henry, C. Ayache, L. Forro, H. Noel, M. Potel, P. Gougeon and J.C. Levet, 1988, J. Phys. (Paris) Colloq. 49, C8-2119. Rossat-Mignod, J., L.P. Regnantt, D. Vettier, P. Burlet, J.Y. Henry and G. Lapertot, 1991, Physica B 169, 58. Sachdev, S., and J. Ye, 1992, Phys. Rev. Lett. 69, 2411. Sato, H., and K. Maki, 1974, Int. J. Magn. 6, 183. Shamoto, S., M. Sato, J.M. Tranquada, B.J. Sternlieb and G. Shirane, 1993, Phys. Rev. B 48, 13817. Shirane, G., Y. Endoh, R.J. Birgeneau, M.A. Kastner, Y. Hidaka, M. Oda, M. Suzuki and T. Murakami, 1987, Phys. Rev. Lett. 59, 1613. Shirane, G., R.L Birgenean, Y. Endoh, RM. Gehring, M.A. Kastner, K. Kitazawa, H. Kojima, I. Tanaka, T.R. Thurston and K. Yamada, t989, Phys. Rev. Lett. 63, 330. Shraiman, B.I., and E.I. Siggia, 1989, Phys. Rev. Lett. 62, 1564. Si, Q., Y. Zha, K. Levin, J.P. Lu and J.H. Kim, 1993, Phys. Rev. B 47, 9055. Singh, R.R.P., 1989, Phys. Rev. B 39, 9760. Singh, R.R.R, P.A. Fleury, K.B. Lyons and P.E. Sulewski, 1989, Phys. Rev. Lett. 62, 2736. Squires, G.L., 1978, Introduction to the Theory of Thelxnal Neutron Scattering (Cambridge University Press, Cambridge)~

Statt, B.W, L.M. Song and C.E. Bird, 1997, Phys. Rev. B 55, 11122. Stem, R., M. Mali, J. Roos and D. Briukmann, 1995, Phys. Rev. B 52, 15734. Stemlieb, B.J., G.M. Luke, Y.J. Uemura, T.M. Riseman, J.H. Brewer, P.M. Gehring, K. Yamada, Y. Hidaka, T. Murakami, T.R. Thurston and R.J. Birgeneau, 1990, Phys. Rev. B 41, 8866. Stemlieb, B.J., J.M. Tranquada, G. Shirane, S. Shamoto and M. Sato, 1994, Phys. Rev. B 50, 12915. Tanamoto, T., H. Kohno and H. Fukuyama, 1994, J. Phys. Soc. Jpn. 63, 2739. Thnrston, T.R., R.J. Birgeneau, M.A. Kastner, N.W. Preyel; G. Shirane, Y. Fuji, K. Yamada, Y. Endoh, K. Kakurai, M. Matsuda, Y. Hidaka and T. Murakami, 1989, Phys. Rev. B 40, 4585. Tranquada, J.M., 1998, J. Phys. Chem. Solids 59, 2150. Tranquada, J.M., G. Shirane, B. Keimer, S. Shamoto and M. Sato, 1989, Phys. Rev. B 40, 4503. Tranquada, J.M., W.J.L. Buyers, H. Chou, T.E. Mason, M. Sato, S. Shamoto and G. Shirane, 1990, Phys. Rev. Lett. 64, 800. Tyc, S., B.I. Halperin and S. Chakravarty, 1989, 62, 835. Vaknin, D., S.K. Sinha, D.E. Moncton, D.C. Johnston, J.M. Newsam, C.R. Safanya and H.E. King, 1987, Phys. Rev. Lett. 58, 2802. Varma, C.M., RB. Littlewood, S. Schmitt-Riuk, E. Abrahams and A.E. Ruckenstein, 1989, Phys. Rev. Lett. 63, 1996. Yamada, K., K. Kakurai, Y. Endoh, T.R. Thurston, M.A. Kastner, R.J. Birgeneau, G. Shirane, Y. Hidaka and ~E Mm'akami, 1989, Phys. Rev. B 40, 4557. Yamada, K., S. Wakimoto, G. Shirane, C.H. Lee, M.A. Kastner, S. Hosoya, M. Greven, Y. Endoh and R.J. Birgeneau, 1995, Phys. Rev. Lett. 75, 1626. Yamada, K., C.H. Lee, K. Kurahashi, J. Wada, S. Wakimoto, S. Ueki, H. Kimura, Y. Endoh, S. Hosoya, G. Shirane, R.J. Birgeneau, M. Greven, M.A. Kastner and Y.J. Kim, 1998, Phys. Rev. B 57, 6165. Zaanen, J., and O. Gunnarsson, 1989, Phys. Rev. B 40, 7391. Zha, Y., K. Levin and Q. Si, 1993, Phys. Rev. B 47, 9124. Zhang, S.-C., 1997, Science 275, 1089.

Handbook on the Physics and Chemistry of Rare Earths Vol. 31 edited by K.A. Gsehneidner,Jr., L. Eyring and M.B. Maple 2001 Elsevier Science B. g

Chapter 199 NEUTRON SCATTERING STUDIES OF LANTHANIDE MAGNETIC ORDERING J.W. L Y N N N I S T Center f o r Neutron Research, National Institute o f Standards and Technology, Gaithersburg, M D 20899, USA and Center f o r Superconductivity Research, Department o f Physics, University o f Maryland, College Park, M D 20742, USA S. S K A N T H A K U M A R

Chemistry Division, Argonne National Laboratory, Argonne, IL 60439, USA

Contents

1. 2. 3. 4.

List of symbols Introduction General trends Single-layer systems Multilayered systems 4.1. Cu spin ordering 4.2. Lanthanide ordering in RBa2Cu307 4.3. R2Ba4CusO16 and RzBa4Cu7Ol5 systems - ideal 2D magnetism

315 316 318 318 330 332 333

4.4. Oxygen dependence of the lanthanide ordering in RBazCu306+ x 4.5. Pb2Sr2R 1 xCa, Cu308 5. The special case of Pr 6. Lanthanide spin dynamics 7. Overview and future directions Acknowledgments References

337

List of symbols I(z-)

neutron intensity at reciprocal lattice vector z"

J J J~,b

total angular momentum

1-2-3

RBa2Cu306+ x

2-4-8 2-4-7 2-1-4 2D 3D A

R2Ba4Cn8016 R2Ba4Cu7OI5 R2 xCe, CuO4 two-dimensional three-dimensional average size of domain in the a-b layer

a, b, c

lattice parameters instrumental constant for the calculation of neutron intensity

k

magnetic structure factor magnetic form factor Miller indices of Bragg peaks

qm R RKKY

C FM f(T)

h, k, 1

exchange energy exchange interaction energy in the

a-b plane Jc

exchange interaction energy along the c-direction Boltzmann constant unit vector along the spin direction

315

antiferromagnetic propagation vector Lanthanide ion Ruderman-Kittel Kasuya-Yosida

339 340 340 344 345 347 347

316 S S~ T T~ TN

t

J.W. LYNNand S. SKANTHAKUMAR total spin momentum spin operator at site i temperature superconductingtransition temperature antiferromagneticordering (N6el) temperature reduced temperature (1 - T/TN)

Y~e2_)2 neutron-electroncoupling constant, mc2J -0.27x10 -12cm

3. #B (#z)

London penetration length Bohr magneton thermal averageof the ordered magnetic moment superconducting coherence length reciprocal lattice or scatteringvector unit vector along the reciprocal lattice vector r

1. Introduction

The effects of magnetic impurities and the possibility of magnetic ordering in superconductors have had a rich and interesting history (see the reviews by Maple 1976 and Fischer and Maple 1983). Magnetic impurities substituted into a superconductor were found to quickly suppress superconductivity due to the strong spin scattering that disrupts the Cooper pairs. Typically ~1% substitution was enough to completely extinguish the superconducting state, and such a low concentration of magnetic moments precludes the possibility of cooperative magnetic states forming and competing with the superconducting order parameter. The first exception to this behavior was realized for the (Cel xRx)Ru2 system, where over 30% of non-magnetic Ce 4+ could be replaced by the magnetic heavy lanthanides before superconductivity was suppressed. Strong ferromagnetic correlations were found to develop in the superconducting state, but no long-range order was present. The first examples of true long-range magnetic order coexisting with superconductivity were provided by the ternary Chevrel-phase superconductors (RMo6S8) and related (RRh4B4) compounds (Fischer and Maple 1983). In these materials there is a separate, fully occupied lanthanide sublattice. The fact that these materials were superconducting at all implied that the magnetic ions and the superconducting electrons belonged to different, "isolated" sublattices, and thereby the conventional Abrikosov Gorkov (1961) spin-depairing mechanism was suppressed. The magnetic ordering temperatures are all low, ~1 K, and thus it was argued that electromagnetic (dipolar) interactions should dominate the energetics of the magnetic system. For materials where these interactions favor antiferromagnetism the magnetization averages to zero on the length scale of a unit cell (a), which results in a weak influence on the superconducting state (a << ~, )0- This is the most prevalent case found in nature, and apart from a few anomalies in properties such as the upper critical field, the antiferromagnetic order and superconductivity were found to readily accommodate one another. In the rare and more interesting situation where the magnetic interactions are ferromagnetic, there is strong coupling to the superconducting state that originates from the internally generated magnetic field. The competition with the superconducting order parameter gives rise to long wavelength oscillatory magnetic states and/or reentrant superconductivity; the neutron work has been

LANTHANIDEMAGNETICORDER:NEUTRONSCATTERINGSTUDIES

317

reviewed by Thomlinson et al. (1983), with some subsequent work (Lynn et al. 1984). The studies of these materials contributed greatly to our understanding of these two competing phenomena, but the possible role of exchange interactions in these systems, and the related question of how the dipolar interaction alone could be responsible for the antiferromagnetism in some materials while others (with the same crystal structure) are ferromagnets, remained unexplained. The cuprate superconductors offer new and interesting perspectives into our understanding of"magnetic superconductors" for a number of reasons. In the materials typified by RBa2Cu306+x (1-2-3) and R2Ba4CusO16 (2-4-8), for example, the R ions appear to be electronically isolated from the Cu-O layers where the Cooper pairs form, similar to the ternary superconductors, and the low lanthanide ordering temperatures (~1 K) suggested that the cuprate superconductors were again prototypical "magnetic superconductors". They were also interesting because the layered crystal structure, with c ~ 3a, rendered them naturally two-dimensional (2D) in nature, and indeed some of the best 2D magnets known belong to this class of materials (see the review by Lynn 1992). However, there are also examples where the magnetic ordering temperature is much too high to be explained by dipolar interactions, and it has become clear that R - R exchange interactions actually must play a dominant role in the magnetism, as is also clearly the case for the new RNizB2C class of superconductors (Lynn et al. 1997, Stassis and Goldman 1997). In contrast to the ternary superconductors, in the cuprates there is no clear separation of the lanthanide sublattice from the superconducting electrons. Moreover, one of the most interesting aspects of the cuprates concerns the magnetism associated with the Cu ions, which is the same sublattice where the superconducting pairing occurs. The undoped cuprates are antiferromagnetic insulators where the S = gi Cu spins order at high temperatures, typically near or above room temperature. The in-plane Cu exchange interactions are much stronger than along the c-axis, and thus again the magnetism is twodimensional in nature. With doping, the materials lose the Cu long-range magnetic order and become high-To superconductors. However, the Cu moments and energetics are still present, and the essential role these quantum spin fluctuations play in the superconducting state is reviewed in ch. 198. In the present chapter we review the neutron scattering investigations of the magnetic structures of the lanthahides in the cuprates. We start by discussing the tetragonal, single Cu-O layer electron-doped materials typified by (Nd-Ce)2CuO4. These are the simplest and best understood from a number of standpoints, but the lanthanide ions are clearly exchange coupled in these materials, and there is also an important coupling between the Cu and R moments. This is particularly evident when the Cu spins exhibit longrange order, and we will therefore briefly describe the nature of the Cu order as well, and its interaction with the lanthanide moments. We then discuss the 1-2-3- and 2-4-8type layered materials, which often exhibit prototypical 2D behavior. Next we will turn to the behavior of the Pr ion in the cuprates, which is fundamentally different. In this case the 4f electrons hybridize with the conduction electrons, which results in Pr having the highest lanthanide ordering temperatures in this class of materials. This hybridization also typically extinguishes the superconductivity. Finally, we will summarize our overall

318

J.W. LYNN and S. SKANTHAKUMAR

understanding of the lanthanide ordering in these materials, and briefly discuss future directions. 2. General trends Before discussing the details of these materials, there are some general trends that should be noted. For the Cu spins, the central feature that controls many aspects of all the oxide materials is the strong copper-oxygen bonding, which results in a layered Cu-O crystal structure. In the undoped "parent" materials this strong bonding leads to an electrically insulating antiferromagnetic ground state. The exchange interactions within the layers are much stronger than between the layers, and typically an order-of-magnitude more energetic than the lattice dynamics. The associated spin dynamics and magnetic ordering of the Cu ions are thus driven by this two-dimensional nature. With electronic doping, long-range antiferromagnetic order for the Cu is suppressed as metallic behavior and then superconductivity appears, but strong antiferromagnetic spin correlations still persist in this regime. It is this large magnetic energy scale that is associated with the high superconducting transition temperature and exotic pairing, and these aspects are reviewed by Mason in ch. 198. For the lanthanide magnetism, the overall behavior is quite different. In the singlelayer R2CuO4 type materials the lanthanide exchange interactions are three-dimensional in nature, and typically the Cu and R systems exhibit relatively strong coupling. For the multi-layered materials, on the other hand, the distance along the stacking axis becomes quite large, and this physical separation renders the ordering two-dimensional in nature, but for very different reasons than for the Cu system. The R-Cu interactions also tend to be weaker and the lanthanide sublattice appears to be relatively isolated from the rest of the system, making them prototypical 2D magnets with low ordering temperatures. In all of these systems, for both the Cu magnetic order as well as the lanthanide magnetic order, the spin structures are relatively simple commensurate antiferromagnetic configurations. In particular, no materials have been discovered yet that exhibit a net ferromagnetic component, with its associated macroscopic magnetization. By commensurate we mean that the antiferromagnetic unit cell is a simple integer multiple of the chemical unit cell. Within the Cu-O planes, for example, the nearest-neighbor spins have always been found to be antiparallel, and the spins always point along a common direction (i.e., they are collinear) within the a - b plane. However, there can be the complication that some of the structures are noncollinear, by which we mean that the direction of the spins between the layers is not necessarily along the same crystallographic direction. The added complication of having both Cu and R moments can then make the magnetism of these cuprates quite rich and interesting. 3. Single-layer systems We start our discussion with the magnetic ordering in the electron-type R2CuO4 (2-1-4) materials. The basic crystal structure is tetragonal I4/mmm (T r phase) in which the Cu

LANTHANIDEMAGNETICORDER:NEUTRONSCATTERINGSTUDIES

319

R2_xCexCuO4

L@

~ •

@ R,Ce

~

• Cu

©0

y

Fig. 1. T' tetragonalcrystalstructureof the singleCu-O layerR2CuO 4 system.

spins occupy a body-centered tetragonal (bct) lattice (a ~ 3.91 A, c ~ 11.9A) as shown in fig. 1. The Cu-O bonding within the a-b plane results in strong exchange interactions within these layers, while the rare-earth ions form layers between these Cu-O planes. The Cu spins in the undoped ("parent") material are antiferromagnetically ordered and the material is insulating, while ~7.5% Ce or Th substituted onto the R site (R = Pr, Nd, Sm, Eu) gives the optimal doping for superconductivity, with Tc ~ 25 K. The phase diagrams as a function of doping for all of these cuprate materials are reviewed elsewhere in this Handbook (Maple 2000, Elschner and Loidl 2000). For the Cu spins the c-axis coupling is thus much weaker than the coupling in the a-b plane, giving rise to both two-dimensional magnetic properties and highly anisotropic superconducting behavior. Note, however, that the exchange interactions for both the Cu and R moments, as well as the superconducting order parameter, must be mediated through magnetically active layers. When a material undergoes a transition from a paramagnet to a magnetically ordered state, new Bragg peaks develop that are associated with the long-range magnetic order. The positions of these new magnetic reciprocal lattice points ~ identify the magnetic configuration of the spins, while the size and spin direction of the moments can be determined from the intensities. For a simple collinear magnetic structure the neutron intensity is given by (Bacon 1975) (ye2 ~ 2

= c \2mcRj

2

2 Ifel

2

l1

-

(.~ ~¢)2) •

,

(1)

where C is an instrumental constant, the quantity in large parentheses is the neutronelectron coupling constant (-0.27x 10-t2 cm), (/~z) is the thermal average of the ordered magnetic moment, f ( r ) is the magnetic form factor (the Fourier transform of the atomic magnetization density), FM is the magnetic structure factor, and "~ and M are unit vectors in the direction of the reciprocal lattice vector ~ and the spin direction, respectively. The orientation factor ( 1 - (7. ~t) 2} must be averaged over all possible

320

J.W. LYNN and S. SKANTHAKUMAR . . . .

i

. . . .

a~~'

i

. . . .

~

. . . .

i

" ~

. . . .

i

. . . .

i

. . . .

Sm2CuO4

• 0 0

50

100

150

200

250

Temperature(K)

300

350

Fig. 2. Temperature dependence of the (½,½,0) Cu magnetic Bragg peak in Sm2CuO4. The solid curve is a [cast-squares fit to a simple power law, [ 10(1 T/TN) 2~, with/3=0.30(1) and TN =280(1)K (Skanthakumar eta!. t991).

domains. A similar, although more complicated, expression is obtained if the magnetic structure is noncollinear, but the general behavior in terms of magnetic moment, form factor, and determination of the spin directions is qualitatively the same. Figure 2 shows the intensity of the (g,g,0) I 1 magnetic Bragg peak associated with the Cu ordering in Sm2CuO4 (Skanthakumar et al. 1991). We see from eq. (1) that the intensity is proportional to the square of the ordered moment, which in this case is the sublattice magnetization. The smooth variation with temperature indicates that the ordered moment develops continuously below the N6el temperature of 280 K. However, we note that the intensity does not exhibit the usual saturation at low T typical of a conventional three-dimensional order parameter. The solid curve is a fit to a power law t 2~, where t is the reduced temperature and/3-0.30-4-0.01. This is the expected behavior near the ordering temperature, but it is unusual for a power law to fit the data over such a wide temperature range. This has been observed in other S = g1 C u - O systems, and is thought to originate from the two-dimensional quantum fluctuations present in these highly anisotropic antiferromagnets. The magnetic Bragg peaks observed in the magnetically ordered state can be indexed as (½h, ½k, l) based on the chemical unit cell, where h and k are odd integers and l is any integer. Since the first two Miller indices are half integers, the Cu magnetic unit cell is double the chemical unit cell along the a and b directions while it is the same along c. This Cu magnetic unit cell is the same as that found for the other 2-1-4 systems. The interactions between spins within an a-b plane are antiferromagnetic as already discussed, so that nearest-neighbor spins within a layer are antiparallel, as is the case in all the cuprates. The nearest-neighbor exchange interaction JSi.Sj between layers, on the other hand, is seen to cancel due to the body-centered tetragonal (bct) symmetry, further rendering the net Cu spin interactions 2D in nature. Hence the three-dimensional

LANTHANIDE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES

321

Cu spin structures in R2CuO 4

a

(a)

(b)

(c)

(d)

Fig. 3. The fbur possible Cu spin structures in tetragonal RaCuO4: (a) collinear with qM ± M; (b) noncollinearSm2CuO4 type (c) collinear with qM ]13~; (d) noncollinear-Nd2CuO 4 type.

magnetic structure that is actually realized must be stabilized by higher-order interactions (Yildirim et al. 1996), as we will discuss shortly. The detailed spin structure, which entails assigning a spin direction to each site, turns out to contain an ambiguity; there are two possible descriptions (figs. 3a,b) that may occur for the present case where the crystal structure has tetragonal symmetry, and it is not possible to distinguish between them with neutron diffraction data on a multidomain sample. One possibility is a collinear spin structure as shown in fig. 3a, which is the same structure as observed in orthorhombic La2CuO4 (Vaknin et al. 1987). The antiferromagnetic propagation vector qM that describes this structure is along the [110] direction, while the spin direction M is orthogonal, along [110]. The structure then consists of ferromagnetic sheets in the (110) plane, with the spins in adjacent sheets antiparallel (qM _1_M) and the magnetic symmetry is orthorhombic. A second, completely different possibility is the noncollinear spin assignments shown in fig. 3b. In this structure the spins within each Cu-O plane are again collinear and antiferromagnetically coupled as before, but the spins between adjacent planes are rotated by 90 ° and hence are noncollinear. The spin direction in this structure is either along the [100] axis or along the [010] axis, and the basic magnetic symmetry in this case is tetragonal rather than orthorhombic. These two basic magnetic structures, collinear and noncollinear, are in fact closely related: the noncollinear structure can be obtained from the collinear structure by the coherent addition of two separate domains of the collinear structure. From a scattering point of view, the intensities of the magnetic Bragg peaks from a domain of the collinear structure would be quite different from those for the noncollinear structure. However, generally a material exhibiting the collinear structure would form equal populations of domains, and the intensities from the multiple-domain sample are identical to the intensities for the noncollinear structure. Thus in a zero-field experiment the two types of structures cannot be distinguished, and one must apply a symmetry-breaking magnetic field to identify the correct structure. If we apply the field along the [110] direction for

322

J,W. LYNN and S. SKANTHAKUMAR

R2CuO4

SmaCuO4 25000 0

(lo~)

J

20000

:I

15000 4J E B ~0000 © 0 4J 5000 @ Z 0

I

oR

a

2

(a)

3

4

Tsmpspstup8

5

~

(K)

'~

7

J

_ (b) R =sm

J

f (c) R=Gd,em

Fig. 4. (a) Temperature dependence of (1,0,1) Sm magnetic Bragg peak intensity (Smnarlin et al. 1992). (b) Sm magnetic structure. (c) Magnetic structure for Gd and Cm.

example, one can see from fig. 3 that the behavior of the spin system will be quite different depending on whether the structure is collinear, where spins are parallel or antiparallel to the field direction, or noncollinear, where the spins initially are at 45 ° to the field direction. The results for Nd2CuO4 (Skanthakumar et al. 1993a,b) demonstrate that the correct structure is the noncollinear one (Skanthakumar et al. 1989, Petitgrand et al. 1990), and the noncollinear structure appears also to be correct for Sm2CuO4 (Skanthakumar et al. 1993a), Pr2euO4 (Sumarlin et al. 1995), and E u 2 C u O 4 (Chattopadhyay et al. 1994b). For the lanthanide ordering, the simplest example is provided by Sm2CuO4, which also turns out to be an interesting example of the interaction between the lanthanide magnetism and superconductivity. All observed Sm magnetic peaks were found to coincide with nuclear peaks, and they can be indexed as (h,k,l) based on the chemical unit cell, where all h, k and l are integers and h + k + l = even integer. The intensity of the [101] magnetic Bragg peak is shown in fig. 4a, where we find a N6el temperature of 5.95 K (Sumarlin et al. 1992). The magnetic transition is clearly very sharp, and indicates that the lanthanide and Cu magnetic structures do not interact significantly with each other. The fact that the Miller indices are integers signifies that the magnetic unit cell is identical to the chemical unit cell. The magnetic structure for the Sm moments is shown in fig. 4b, where we see that the moments form ferromagnetic sheets in the a-b plane, that are coupled antiferromagnetically along the c-axis, with the spin direction also along c. The Sm magnetic structure, which is in the I4/m~mm magnetic space group, then has a completely different symmetry than the Cu structure, and the spin directions are orthogonal as well, so that the two magnetic phases in this material are independent of each other to a very good approximation. The ordered magnetic moment of the Sm ion is only 0.37(3)#m which is quite small compared to the heavy lanthanides. With the high ordering temperature o f - 6 K , it is quite clear that the energetics must be dominated by exchange interactions, with the dipolar interactions playing only a very minor role. In the doped superconducting

L A N T H A N I D E M A G N E T I C ORDER: N E U T R O N S C A T T E R I N G S T U D I E S

Nd2CuO4 • ,.

, ....

~

,

,

323

Pr2CuO4 .

.

.

.

,

Z

(3/23/23)

30

80

130 180 230 280

Temperature (K)

)

50

/

100 150 200 250 300

Temperature (K)

Fig. 5. (left) Temperature dependence of five Nd2CuO4 magnetic Bragg peaks (Skanthakumar et al. 1989, 1999). (right) Temperaturedependence of fivePr2CuO4magneticBraggpeaks. The solid curves are fits to the theory (Sachidanandam et al. 1997); the dashed curves on the right show the contributionof the Cu spins to the intensities.

system, the dilution of the Sm sublattice smoothly decreases the N6el temperature to ~4.9 K, with little other effect on the magnetism. In the superconducting state (Tc = 23 K), but above the Sm antiferromagnetic ordering temperature, the Cu-O layers contain a mirror (m) symmetry, which the superconducting wave function must accommodate. When the Sm ions order magnetically, on the other hand, then the Cu-O layers contain an antimirror symmetry (rn~), and the superconducting wave function must then accommodate this change in symmetry. Viewed along the c-axis, the system consists of alternating superconducting and ferromagnetic layers, which should require a change of sign of the superconducting order parameter in adjacent C u - O layers. This "g-phase" model is predicted to substantially affect the superconducting state below the magnetic ordering temperature (Andreev et al. 1990), and this is an active area of interest both in this natural layered system and in artificially layered films. The nature of the lanthanide moments and magnetic ordering can also have important consequences for the Cu magnetic structure, and vice versa, as we will now discuss. The column on the left of fig. 5 shows the temperature dependence of five magnetic

324

J.W. LYNN and S. SKANTHAKUMAR

lhCuO4 Aw"

"S

i

•

(b) Nd ( p h a s e - I I )

R

© Cu

.or

(a) Nd (phase-I & III)

"3=

L a

(c)Pr

Fig. 6. Noncollinear spin structures for (a) Nd in phase I and III, (b) Nd in phase II, and (c) Pr.

Bragg peaks for Nd2CuO4. Five magnetic transitions have been identified (Skanthakumar et al. 1989, Endoh et al. 1989, Lynn et al. 1990, Matsuda et al. 1990). The initial ordering of the Cu spins occurs at a N6el temperature of 276 K for this sample, where all the (g,~,/)-type 11 reflections increase in intensity with decreasing temperature in phase I (75 K~< T ~<276K), and the noncollinear Cu spin structure is that shown in fig. 3d. We remark that the Cu ordering temperature is quite sensitive to small changes in the oxygen content, with TN varying from 245 K to 276K (Skanthakumar 1993). At 75 K the intensities of the odd-integral (odd-l) peaks suddenly drop, while the even-integral peaks increase abruptly in intensity. This indicates that a spin-reorientation transition for the Cu spins has occurred, and the new noncollinear Cu spin structure is shown in fig. 3b. At 30 K another abrupt spin reorientation takes place, where the spins rotate back to the original spin sense as indicated by the negligible intensity for T < 30 K of the (1,½,0) peak, which has nonzero intensity only in Phase II (30 K ~< T ~<75 K). Note that this is the same peak that has the strongest intensity for Sm2CuO4, and it turns out that these transitions are the result of a spin reorientation from one noncollinear structure to the other, and then back again, as shown in fig. 6. Similar types of spin-reorientation transitions have been observed in some other related compounds (e.g. La2CoO4), but those transitions are accompanied by structural phase transitions (Yamada et al. 1989). In N d 2 C u O 4 , there is no evidence for structural phase transitions accompanying the magnetic order, and hence the origin of these spin-reorientation transitions must be different. Finally, the Nd ions order antiferromagnetically in the vicinity of 1.5 K, also in a noncollinear spin arrangement, and the magnetic symmetry of the Nd spins is identical to that of Cu spins. In addition to all these transitions, there is a fifth transition of a continuous nature at 0.15 K, which is due to an induced ordering of the nuclear spins through the hyperfine interactions. Such a nuclear spin ordering was originally introduced to explain the low-temperature data of both Nd2CuO4 and Sm2CuO4, and has been recently investigated thoroughly in Nd2CuO4 (Chattopadhyay and Siemensmeyer 1995).

LANTHANIDE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES 0.10

'

'

'

'

I

'

'

'

'

I

'

'

'

'

I

. . . .

I

. . . .

f

'

'

'

325

'

~ ( i e ~, )h<S> =v (h°,= = 0.139meVatT=OK) 0.08

r2CuO4

E 0.06 ,e

"~

0.04

0.02

0.00

0

50

100

150

200

250

300

1.5 1.5

(b) Nd2CuO4

''1 ....

I ....

i ....

I ....

~1.0

E .~ 0.5

0.5 0 .0.1. - .

~

.

5

.3

I

,

.i

i

,

,

~ 20

0.0 - , , , , 0

0

t

1

~

1

. .1. .~. .,

~

30

~

"-

40 t

50

i . . . . i . . . . -T ,', ,', ¢, ,=, ,'I ,,,--'7~ 50 100 150 200 250 300 Temperature (K)

Fig. 7. Calculated polarization of the (a) Pr and (b) Nd moments in R2CuO4, via the interaction with the Cu ordered moments, as a function of temperature (Saehidanandam et al. 1997).

Below 3 0 K (phase III) the intensities evolve in a rather complicated way, and this turns out to be caused by the development o f substantial (staggered) moments on the N d sites as indicated in fig. 7b. Note in particular that the (g, 1 g,3) 1 peak varies substantially in intensity at lower temperatures (fig. 5), and this is the strongest peak associated with the N d ordering at low temperature (Lynn et al. 1990, Matsuda et al. 1990). The antiferromagnetically ordered Cu sublattice induces the N d spins to order in a staggered arrangement. There are four Cu nearest-neighbor spins, one next-nearest-neighbor and four third-neighbors, for each N d ion. The nearest-neighbor C u - N d interactions cancel exactly due to the tetragonal crystal symmetry. However, next-nearest-neighbor and third-neighbor interactions do not cancel, but instead produce a field that polarizes the N d sublattice antiferromagnetically. This polarization field turns out to have the same symmetry as when the N d ions order spontaneously, and thus the Nd ions are quite

326

LW. L Y N N a n d S. S K A N T H A K U M A R

susceptible even at elevated temperatures, and a net sublattice magnetization of the Nd ions is observed in all the ordered phases (fig. 7b). An induced Nd moment has also been observed in resonant X-ray scattering measurements (Hill et al. 1995). Magnetic structures for the Nd spins along with the Cu spins are shown in figs. 6a,b for all phases. Note that the induced moments at elevated temperatures (over 100 K), and the spontaneous order at low T, are noncollinear structures. Note also that the intensities of the (2,~7,1) 1 1 1 1 and (g,g,2) peaks increase with decreasing temperature in phase II, while tile intensities of the (g,g,0) 1 1 l 1 peaks decrease. This behavior is reversed in and (g,g,3) phase III, and these observations indicate that the Nd and Cu spins along the c-direction are coupled antiparallel in phase II, while they are parallel in phases I and III (figs. 6a,b). The temperature dependence of the other magnetic peaks as well as detailed refinements of the intensities (Skanthakumar 1993, Petitgrand et al. 1992) confirm this type of N d - e u coupling. The refined magnetic moment for Nd is 1.44, 0.04, and 0.014/~B at 0.4, 50 and 80 K, respectively. This shows in a direct way that the Nd and Cu spin systems are coupled, and because of the high temperatures involved this again must be exchange coupling rather than dipolar coupling. This has important implications for the superconductivity since the Nd moments and Nd ordering occur in the superconducting system; the conventional wisdom would say that the Abrikosov-Gorkov spin-depairing mechanism would surely destroy any chance for superconductivity in the material. Thus even if we were to discard the fact that there are Cu spin fluctuations in the superconducting phase of these materials, we still have to deal with the fact that the lanthanide moments are there, and are directly coupled through the superconducting planes. Fhe properties of both the Cu and lanthanide magnetic structures, ordered moments, and ordering temperatures are summarized in table 1, along with some related systems. For Pr2CuO4 the Cu spin structure (Cox et al. 1989, Allenspach et al. 1989, Matsuda et al. 1990, Sumarlin et al. 1995) is the same as that (fig. 3d) observed in Nd2CuO4 at high temperature. The temperature dependence of five typical magnetic peaks is shown in the right-hand column of fig. 5, where we see that the nonmonotonic behavior again indicates that R C u coupling is also present in Pr2CuO4. The spin structure is shown in fig. 6c, and the (small) induced Pr moment (obtained from theory discussed below) is shown in fig. 7a. We see that the Pr and Cu spins are coupled antiferromagnetically along the c-direction, in contrast to Nd2CuO4. Measurements under high pressure in Pr2CuO4 suggest spin-reorientation transitions (Katano et al. 1993) similar to Nd2CuO4, but the Nd system is the only one to exhibit these under ambient pressure. For Eu2CuO4, the Eu 3÷ ion has a non-magnetic ground state, and therefore only Cu magnetic ordering is observed in this system (Chattopadhyay et al. 1994b, Gukasov et al. 1992). In Gd2CuO4 (Chattopadhyay et al. 1991b, 1992) and Cm2CuO4 (Soderholm et al. 1999), the lanthanide and Cu spin structures are different. Gd and Cm order ferromagnetically in the a - b plane and are coupled antiferromagnetically along c like Sm2CuO4, with relatively high ordering temperatures. However, the spin direction is in the a - b plane, and the easy direction within this plane cannot be obtained by (zero-field) neutron data. Note that superconductivity does not occur in these two compounds for any Ce doping. Finally, we note that there

LANTHANIDE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES

327

Table 1 Magnetic properties of R2CuO4". The Cu ordering temperature, TN(Cu), and ordered moment, #c,, change with oxygen concentration. Here we give the highest observed T N and the corresponding/Zcu Material

LaaCuO4 b PhCuO4 Nd2CuO4 Sm2CuO 4 Eu2CuO 4 GdzCuO 4 CmaCuO 4 Sr2CuO2CI2 CaaCuO2C12

TN(Cn) (K)

#c, (#B)

325 284 276 280 265 285

0.5 0.40(2) 0.46(5) 0.38(4) 0.4(1)

251 247

0.34(4) 0.25(10)

Magnetic TN(R) (K) structure, Cu fig. fig. fig. fig. fig. fig. d fig. e

3a 3d 3d c 3b 3b 3c or d

Induced ~1.5 5.95(5) 6.4 25

#R (/~)

Magnetic structure, R

0.080(5) 1.44(5) 0.37(3) J -- 0 7 5.0(5)

fig. 6c fig. 6a fig. 4b fig. 4c fig. 4C

3a or b

Ref.

1 2 5 4,6-14 8,15-17 18,19 20,21 22 23,24 25

La, Eu, Sr and Ca ions do not a carry magnetic moment. The ground state of Pr is a nonmagnetic singlet and only a small induced moment is observed for Pr due to exchange mixing. b Only La2CuO 4 has an orthorhombic (Cmca) crystal structure; all others are tetragonat (I4/mmm). However, recently orthorhombic structural distortions have been reported in Eu2CuO 4 and Gd2CuO4 (Vigoureux etal. 1997a,b). c Spin reorientation occurs at 75 K and 30 K. Between these two temperatures the correct structure is represented by fig. 3b. d Cu ordering in Cm2CuO 4 has not been studied yet, but the Cu spin structure is not expected to be different from other R2CnO 4 compounds. e Cu spin structure is different from all other related materials. In this system, the magnetic unit cell is doubled along the c-direction.

References [1] Vaknin et al. (1987); review: Lynn (1990) [2] Cox et al. (1989) [3] Allenspach et al. (1989) [4] Matsuda et al. (1990) [5] Sumarlin etal. (1995) [6] Skanthakumar et al. (1989) [7] Skanthakumar et al. (1993a) [8] Skanthakumar et al. (1993b) [9] Lynn et al. (1990) [10] Endoh et al. (1989) [11] Chattopadhyay et al. (1991a) [12] Akimitsu et al. (1989) [13] Rosseinsky etal. (1989)

[14] [15] [16] [t7] [18] [19] [20] [21] [22] [23] [24] [25]

Petitgrand et al. (1990) Skanthakumar et al. (1991) Skanthakumar (1993) Sumarlin etal. (1992) Gukasov et al. (1992) Chattopadhyay et al. (1994b) Chattopadhyay et al. (1991b) Chattopadhyay et al. (1992) Soderholm et al. (1999) Vakniu et al. (1990) Yildirim et al. (1998) Vaknin et al. (1997)

is a n i n d i c a t i o n f o r a n i n d u c e d G d m o m e n t d u e to t h e p o l a r i z a t i o n b y t h e C u s u b l a t t i c e i n G d 2 C u O 4 ( C h a t t o p a d h y a y et al. 1 9 9 4 a ) , a n d i n p r i n c i p l e t h e C u s u b l a t t i c e p o l a r i z e s r a r e - e a r t h s p i n s i n all t h e s e c o m p o u n d s . S u p e r c o n d u c t i v i t y d o e s n o t o c c u r i n a n y o f t h e other R2CuO4 (R = Tb-Lu) systems, and no neutron diffraction experiments have been r e p o r t e d yet.

328

J.W. LYNNand S. SKANTHAKUMAR

The effects of Ce doping on the magnetic ordering of R and Cu in R2_xCexCuO4have been studied for the Nd and Pr systems (Lynn et al. 1990, Skanthakumar et al. 1992, 1999, Skanthakumar 1993, Thurston et al. 1990, Rosseinsky et al. 1991, Zobkalo et al. 1991). Taking Ndz_xCexCuO4 as an example, for small Ce concentrations (x < 0.14) the same non-collinear antiferromagnetic order (fig. 6) for both the Cu and Nd spins is observed. Both the average ordered Cu moment and TN decrease with increasing Ce concentration x, and no long-range order is observed at optimal doping of x = 0.15 (7.5% substitution of Ce for Nd). The Cu spin-reorientation transitions also broaden rapidly with increasing x, while the fraction of Cu spins that participate in the spin reorientation also appears to decrease with increasing x. On the other hand, the Nd ordering transition, which is smeared due to the strong Cu-Nd coupling in Nd2CuO4, becomes sharper with increasing x. The temperature dependence of the (g,~,3) i J magnetic peak for crystals with various Ce concentrations is shown in fig. 8a. Since the ordered Cu moment decreases with increasing x, the strength of the Nd-Cu coupling and the associated induced moment also decrease with increasing x, allowing the transition to become sharper. To see this effect directly, the temperature dependence of the ratio between the ordered moments of Nd and Cu for various Ce concentrations is shown in fig. 8b. In the temperature regime where the dominant contribution to the Nd ordered moment is via the Nd-Cu interaction, these temperature-dependent moment ratios fall on a universal curve since the induced Nd moment is proportional to the ordered Cu moment. The proportionality constant depends on the magnetic susceptibility of the Nd ions, and we see that the Nd contribution to the magnetic susceptibility changes only marginally with cerium doping. At lower temperatures (< 1.5 K), the ordered Nd moment develops a contribution from the Nd-Nd interaction, and the ratios for different Ce concentrations depart fi'om the universal curve. Nuclear spin ordering of Nd also contributes to the increase in intensity below ~0.4 K. In the superconducting sample (x =0.15), there is no indication for long-range Cu magnetic ordering. However, a transition to long-range antiferromagnetic order of Nd occurs at 1.2 K, with an ordered moment of 0.85/~m and this transition is relatively sharp as there is no ordered Cu moment to induce Nd ordering at higher temperatures. There is no indication for long-range ordering of either Cu or Nd in crystals with higher cerium concentrations (x > 0.17). However, at low temperatures, broad magnetic peaks with weak intensities due to short-range Nd order are observed. The intensities of these peaks are found to increase with decreasing temperature, analogous to an order parameter, while the widths are approximately temperature independent. This indicates that for higher cerium concentrations the domains are small (~150 A). Oxygen plays a crucial role in both the magnetic and transport properties of the high temperature superconductors. In the electron-doped Nd2 xCexCuO4 system the superconducting properties were found to be very sensitive to both cerium and oxygen concentrations. In fact, even in optimally doped Ndl.85 Ce0.15CuO4, superconductivity can be achieved only by removing a small amount of oxygen from as-grown samples (Radaelli et al. 1994) and not by doping more Ce. The overall effects of oxygen reduction on the Cu magnetic ordering are similar to that of increasing the cerium concentration in

LANTHAN1DE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES Nd2_xCexCuO i

329

4

,

1.0

(1/2,1/2,3)

(a)

°~,,~ r.#3

0.8

x=0.0

0.6~

o

kk 0.4 E

N

•

"~ • •

°°

x = 0.025

o x = 0.125

A t~

• ~,, o

-I

•

L.

0.2 Z

oo F ,

. ~ o

4 o

• o

i

°0

i

....

....

3

~, . . . . . . . . . .

° ~ o ~

4

5

6

\

7

(b)

% o

x=0.0

4

• o

x = 0.025 x = 0.125

::~ 2 o A

1

•

0

1

2

3 Temperature

4

5

o

•

6

7

(K)

Fig. 8. (a) Temperature dependence of the (½,½,3)magnetic Bragg peak for Nd2_yCe~.CuO4 crystals with various Ce concentrations. (b) Temperature dependence of the ratio between the ordered moments of Nd and Ca. At higher temperatures, the ordered (induced) Nd moment is proportional to the Cu moment, and thus the curves for various crystals overlap. At lower temperatures (below 1.5 K), the Nd ions spontaneously order (Skanthakumar 1993, Skanthakumar et al. 1999).

Nd2 xCexCuO4 (Skanthakumar 1993, Matsuda et al. 1992). In Ndl.ssCe0.15CuO4, the observed magnetic intensities due to N d ordering in as-grown samples were found to be larger than those o f the deoxygenated (superconducting) samples. In addition, magnetic ordering o f Cu and, therefore, the N d - C u coupling, are observed in as-grown samples, while they disappear when the excess oxygen is removed. As mentioned earlier, the linear exchange interaction between Cu layers cancels due to the bct symmetry, and the actual spin structure is then a result o f a delicate balance o f superexchange, spin-orbit, and Coulomb exchange interactions (Yildirim et al. 1994, 1996). In the absence o f the R ions, quantum fluctuations would be expected to yield the collinear magnetic structure (Yildirim 1999) and this may indeed be the case for the Sr2CuO2C12 material (Yildirim et al. 1998). However, this leaves unexplained the

330

J.V~ LYNNand S. SKANTHAKUMAR

observation of the noncollinear structures in the R2CuO4, and quantum fluctuations certainly cannot explain the series of spin rotations that occur in Nd2CuO4. It is thus clear that the lanthanide exchange interactions and crystal-field anisotropies must play an essential role in determining the three-dimensional magnetic structures realized in these systems, and the correlation between lanthanide magnetism and superconductivity. Considerable progress in properly taking into account the lanthanide exchange interactions and crystal field anisotropies has been achieved recently (Yildirim et al. 1994, Sachidanandam et al. 1997). The theory includes various novel anisotropic magnetic interactions in the cuprates that have a microscopic origin, along with R-R, R-Cu, and Cu-Cu exchange interactions. The calculations show that due to the exchange field acting on the lanthanide ion coupled with the crystalline electric field interactions, there is a strong single-ion anisotropy that orients both the Cu and R=Pr, Nd spins along the [100]-type directions, into the observed noncollinear spin arrangement. The model also successfully predicts the consecutive spin flop transitions observed in Nd2CuO4, and the calculations (solid curves) shown in fig. 5 are in excellent agreement with the data. The Cu spins order first, and then as the temperature decreases the Nd spins develop a significant polarization. As this polarization develops it disrupts the balance of energies along the c-axis, and this causes an abrupt (first-order) spin reorientation from one noncollinear structure to the other. As the Nd moment increases the energetics shift again and the spins rotate back to their original directions. The polarization of the Nd sublattice, as determined by fitting the model to the intensities of a series of magnetic Bragg peaks, is shown in fig. 7b. We see that there is a substantial induced moment on the Nd below ~100 K, while the Nd exchange orders the moments at ~1.5 K (in the absence of the Cu). This behavior contrasts with the situation for Pr2CuO4, where the Pr crystalfield ground state is a (non-magnetic) singlet (Sumarlin et al. 1995, Stanb and Soderholm 2000). There is then a small moment that is present due to exchange mixing (Matsuda et al. 1990, Sumarlin et al. 1995), and the polarization of this moment as determined from the model is presented in fig. 7a; the excellent fits to the data are shown as solid curves in the right column of fig. 5. The model predicts that there is significant polarization all the way up to TN(CU), but this polarization is too small to cause spin reorientations in Pr2CuO4, in agreement with experiment. Finally, we note that this same type of calculation also predicts that the easy axis lies along [001] for Sm in Sm2CuO4, with no significant coupling of the lanthanide mad Cu spin structures, again in agreement with observation. We conclude then that there is a good overall understanding of the microscopic magnetic interactions, and the consequent magnetic phases and associated spin dynamics, in the single-layer materials.

4. Multilayered systems We have seen in the previous section that the single-layer materials exhibit substantial R-Cu exchange coupling, and they also have modest (for cuprates) superconducting transition temperatures. For the materials that contain multiple Cu-O layers the To values

LANTHANIDEMAGNETICORDER: NEUTRON SCATTERINGSTUDIES

331

are typically much higher because of the stronger coupling between the C u - O layers. For the canonical RBa2Cu3 06 +x (1-2-3) system, there are three copper-oxygen layers of ions in each chemical unit cell, which are stacked along the c-axis. Two of these layers have oxygen ions between the Cu ions in both the a and b crystallographic directions (the Cu-plane layers), and the oxygens cannot be removed. The third Cu layer only has O ions along one axis. This is the so-called "chain layer", and the oxygen concentration can be readily varied in this layer from full occupancy (x = 1) to full depletion (x = 0). Both the magnetic and superconducting properties are very sensitive functions of the oxygen concentration x; in the fully oxygenated case (x = 1) the system is a 90 + K superconductor for all the trivalent lanthanide elements R except Pr, while in the magnetic regime (x ~<0.4) the Cu plane-spins order antiferromagnetically and the system is a Mott insulator. The lanthanide ions are centered between the two Cu plane layers in the 1-2-3 system, but they turn out to be remarkably isolated electronically from the C u - O layers. Indeed all the physical measurements on the RBa2Cu306+x (1-2-3), the R2Ba4CusO16 (2-4-8), and the R2Ba4Cu7OI5 (2-4-7) systems show that the 4f electrons are effectively isolated from the C u - O superconducting layers, as well as from each other. The single exception is for Pr, and we will discuss this unique case separately. The crystal-field splittings are also very important (except for Gd), and for the orthorhombic symmetry appropriate for these materials typically the lanthanide ground state is doubly degenerate for half-integral J (like Dy or Er) or singly degenerate (like Ho) for integral J (and hence non-magnetic). The crystal-field properties have been reviewed by Staub and Soderholm (2000). As far as the nature of the magnetic ordering is concerned for half-integral J , the higher crystalfield levels are at energies much larger than the ordering energy kTN, and consequently do not play an important role in the ordered state. However, the first excited state for integral J is very close to the ground state (except for Tm), and therefore it is possible for the rare-earth ions to order magnetically in these materials as well. In Tm, the singlet ground state is well separated from higher states (~12 meV), and consequently lanthanide magnetic order is not observed in the Tm compounds. For the systems with a magnetic ground state the direct (exchange) overlap of the 4f wave functions is of course out of the question, and the indirect and superexchange interactions turn out to be relatively small. Thus the ordering temperatures are low (typically ~ l K), and dipolar interactions are important in these materials. Hence the lanthanide magnetism is similar in many regards to "conventional" magnetic-superconductor systems. The sublattices for the lanthanide ions are shown in fig. 9 for the 1-2-3 (9a), 2-4-8 (9b), and 2-4-7 (9c) structures. Neglecting the small orthorhombic distortion in the a-b plane, the 1-2-3 lattice is simple tetragonal. The 2-4-8 system then consists of two 1-2-3 unit cells stacked on top of one another, but shifted by b/2. The 2-4-7 system, on the other hand, consists of two 1-2-3 cells stacked directly on top of one another, and then another two layers stacked on top, but shifted by b/2. Hence the 2-4-7 structure consists of 1-2-3 bilayers. The basic ingredient that lowers the effective dimensionality d of this class of magnetic materials to d =2 is that for all three structures the nearestneighbor R distance along the c-axis is three times the nearest-neighbor distance in the a or b directions; typically a ~ b ~ 3 . 9 A , and c ~ 12A. For a dipolar interaction

332

J.W. LYNNand S. SKANTHAKUMAR

(b) c

Fig. 9. Crystal structm-es for the rare-earth ions in the (a) 1-2-3, (b) 2-4-8, and (c) 2-4-7 systems. The twodimensional behavior of the magnetism is due to the large inter-ion spacing along the c-axis compared to the a-b plane (c ~ 3a). In addition, in the 2-4-8 and 2-4-7 structures the net interactions along the c-axis cancel for some magnetic configurations. (E c< l/r 3) the strength o f the coupling along c should be reduced by ~(a/c) 3 ~ 1/27, while either direct or superexchange interactions would be expected to be reduced by even a larger factor. A significant indirect (RKKY) exchange might still be operative over this long distance, but the conduction electron density at the lanthanide site turns out to be small, and in fact the ordering temperatures in both the insulating and superconducting compounds (which in the 1-2-3 materials can be realized by simply varying the oxygen content) are comparable. The final conclusion is that the interaction along c is indeed considerably more than an order o f magnitude smaller than the interactions operating within the a - b plane, and this generates the 2D-like magnetic behavior. This contrasts with the interactions in the 2-1-4 materials, where the nearest-neighbor R distances are all comparable, and there is no 2D-like behavior for the R magnetism. 4.1. Cu spin ordering

The high-temperature antiferromagnetic transition in the 1-2-3 system involves the ordering of the Cu spins in the Cu-plane layers. At x = 0 the N6el temperature TN can exceed 500 K, and then TN monotonically decreases to zero at x ~ 0.4. The spins within the a - b plane are antiferromagnetically aligned, as is always the case, with the spin direction also in the a - b plane. The Cu ions in adjacent layers are located immediately above and below each other along the c-axis, and the exchange interaction along the c-axis is also antiferromagnetic. In particular, the Cu lattice is not body-centered as in the singlelayer materials, and hence there is no cancellation o f the exchange interactions between layers. Nevertheless, the overall behavior is still quite 2D in nature because o f the strong in-plane Cu-O bonding. It is also possible to dope the chain layers in the 1-2-3 material electronically, either from adjacent sites or directly on the Cu chains, and this can cause a moment to develop on the Cu chain sites, which can also order. This is not a problem for

LANTHANIDE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES

333

the heavy lanthanides, but is an important difficulty encountered in the Pr system, and to a lesser extent in Nd 1-2-3. This coupling breaks the symmetry that the lanthanide site enjoys when only the Cu-plane layers are magnetic, and can consequently influence the lanthanide ordering and R-Cu coupling. We will return briefly to this problem when we review the Pr ordering. For the 2-4-8 materials we note that it is not possible to vary the oxygen content significantly, while for the 2-4-7 systems it can be varied from 14 to 15. However, there is no long-range Cu order to consider in these superconducting systems. 4.2. Lanthanide ordering in RBa2Cu307 The R - R interactions in these superconductors are all quite weak, and the antiferromagnetic ordering temperatures are typically ~1 K as already noted. The most thoroughly investigated system is probably ErBa2Cu3Ov, and this material turns out to be an ideal representation of a S _- 1g 2D Ising antiferromagnet. Early investigations were carried out on polycrystalline samples (Lynn et al. 1987), and were followed by detailed studies on single-crystal samples (Lynn et al. 1989, Paul et al. 1989, Chattopadhyay et al. 1988a, 1989). In the top portion of fig. 10 we present a diagram of reciprocal space, showing the positions of the nuclear (crystal structure) and magnetic (spin structure) Bragg peaks which occur at positions like (gh,k,gl). 1 1 If the system were purely 2D in character then we would expect to see rods of magnetic scattering as shown; above the ordering temperature this scattering originates from critical fluctuations, while below TN it originates from low-energy spin waves. The lower portion of fig. 10 shows three scans through the rod, just above the ordering temperature. This strong rod of scattering develops for temperatures ~<1 K, and at the temperature shown the width is solely instrumental in origin. Also shown in the figure is a scan along the rod, which demonstrates that the scattering intensity is essentially uniform. Hence there are no significant correlations between the spins in adjacent a-b layers, even though we are just above TN = 0.62 K as indicated by the measured sublattice magnetization shown in fig. 11. The energetics are clearly 2D in nature, and we thus regard the phase transition at TN as 2D in character. The magnetic structure for Er 1-2-3 is shown in fig. 12a, and consists of chains of spins coupled ferromagnetically along the b-axis, and antiferromagnetic along a and c. The in-plane magnetic structure is found to be identical in the Er 2-4-8 and Er 2-4-7 systems. In the Dy (Goldman et al. 1987, Fischer et al. 1988, Clinton et al. 1991), Nd (Yang et al. 1989, Fischer et al. 1989), and Gd (Paul et al. 1988) 1-2-3 superconductors, on the other hand, nearest-neighbor spins are found to be antiparallel in all three directions, and they exhibit the magnetic structure shown in fig. 12c, with ordering temperatures of 0.9, 0.5, and 2.2 K, respectively. Tb does not form the 1-2-3 structure, but superconducting TbSr2Cu2.7Mo0.307 exhibits the same magnetic structure, with TN = 5.4 K (Li et al. 1997). Tm 1-2-3, on the other hand, possesses a singlet crystal-field ground state, which must be non-magnetic. This illustrates an important point, that the crystal field plays an essential role in these systems (see ch. 194 of Vol. 30 of this Handbook, by Staub and Soderholm). The above four ions (Er, Dy, Gd, and Nd) are half-integral J lanthanide ions, and therefore

334

J.W. LYNN and S. SKANTHAKUMAR

ErBa Cu 0 2

(00l)

3

7

• Nuclear 0 Magnetic

2-d rods

(004) I (003) II (002) IP @

(001) It II,

N

V

(ooo)

A

A

W

W

(loo)

750

T==O.b-TK

(~, 010.5)

7eaF (Q~ O, 0.7)

650 550

550

.450

450

0

-~ea

~_ Q~

7"

~

2.50j~ -

i50 0.41

q O.44

' 0.47

(_~

' 0.50

~5 0. 3

(Q~e O, O. 80---)

0.56

0.59

150 I 0.41 1350

T=0.67K

~_, 0.44

i 0,47

I 0.50

, 0.53

(O. 5, O, Q=)

0.56

0.59

T==0.

ii58

650 550

950 I

D Q

-

aea •

750 ~]~

T=O.67 K

[

-r~ E

250

(h00)(2oo)

;450

750

_~

-

~_ -- ~

250

~

950 [

1500

4i

~

O. 4

'

0.47

'

0.50

'

0.53

o.'5e

0.59

tea 0.3

'

0.5

'

0,7

'

0,9

[

i i

,

i.~

'

i.5

1.7

Fig. 10. (top) Reciprocal space for ErBa2Cu3OT, showing the three-dimensional nuclear and magnetic Bragg peaks, and the two-dimensional rods of scattering. (bottom) Scans through the rod, and along the rod, just above the ordering temperature of 0.618 K, showing that there are strong spin correlations within the a--b planes, but no significant correlations between planes (Lynn et ai. 1989).

LANTHANIDE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES

ErBa Cu 0 2

3

335

7

(~/2,o,±/2)

~..2

TN = 0 . 6 2 K

0 ~E LO F~Z Z 0 I-H

< N t-III Z <

O,B O.B

0.4

0.2

0.2

0,4

0.5

0.8

Temperature

f.0

t,2

T/T N

Fig. 11. Temperature dependence of the sublattice magnetization for ErBa2Cu307. The solid curve is the fit to the data of Onsager's exact solution for the S - ½ , 2D Ising model (Lynn et al. 1989).

RBa2Cu306+x

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 12. Three-dimensional magnetic struc~lres for the RBa2Cu306 ~.~ systems. (a) ErBa2Cu307 (Lynn et al. 1987, 1989, Paul et al. 1989, Chattopadhyay et al. 1988a, Maletta et al. 1990, Clinton et al. 1995). (b) The Er spins in portions of some samples of ErBa2Cu306+ x are coupled ferromagnetically along the c-axis (Chattopadhyay et al. 1988a, 1989, Paul et al. 1989, Clinton et al. 1995). (c) DyBa2Cu307 (Goldman et al. 1987, Fischer et al. 1988, Clinton et al. 1991, 1995), NdBazCu306+ ~ (Yang et al. 1989, Clinton et al. 1995), PrBg2eu306+ x (Li et al. 1989, Gnillaume et aL 1993), GdBa2Cu306.14 (Mook et al. 1988, Paul et al. 1988) and TbSr2Cuz69Mo0307 (Li et al. 1997). (d) DyBa2Cu306 (Guillaume et al. 1994, Clinton et al. 1995), GdBa2Cu306+~ (Chattopadhyay et al. 1988b, Guillaume et al. 1994). (e) Correlations in HoBa2Cu307 (Roessli et al. 1993a). (f) YbBa2Cu?O 7 (Roessli et al. 1992).

in a (low-symmetry) crystal-field environment we expect the ground state to be (at least) doubly degenerate (so that S -_~[ is appropriate). Consequently the lanthanide ions will always carry a moment at low temperatures, and order magnetically. This is also the case for Yb 1-2-3, which orders antiferromagnetically at 0.4K (Roessli et al. 1992). The situation for the case where J is whole-integral, on the other hand, results in a crystal-field ground state that is a (non-magnetic) singlet. Hence Tm 1-2-3 does not order

336

J.W. LYNNand S. SKANTHAKUMAR

(Chattopadhyay et al. 1990). Ho also does not have a conventional magnetic transition, but the large nuclear moment and hyperfine field combine to drive a phase transition at 140mK (Roessli et al. 1993a). It is important to note that the dimensionality of the phase transition is also evident in other measurements such as specific heat and susceptibility, which measurements in fact often precede the neutron experiments and play an essential role in detecting the phase transition and revealing the nature of the magnetism. Specific heat as well as inelastic neutron scattering have also shown that the crystal-field ground state for the Er ion (as well as Dy and Nd) is a doublet, so that an S = 1 (two-state) description is appropriate. Hence this system should be described by an S = g1 2D Ising antiferromagnet. Below the N6el temperature Bragg scattering is observed, and the temperature dependence of the sublattice magnetization (order parameter) is shown in fig. 11. The solid curve in the figure is the exact Onsager (1944) result for the S = 1 2D Ising model, and it is seen that experiment is in excellent accord with this exact theory. Analogous behavior has been observed in DyBa2Cu3Ov (Clinton et al. 1991, 1995). One of the interesting features of "2D" magnetic systems is that generally they will order three-dimensionally even if the interaction in the third direction is very weak (see Lynn 1992 for a review). To understand the origin of this behavior, we designate Jab to be the basic interaction within the a - b plane, which will consist of possible exchange plus dipolar energies, and Jc as the energy (likely dominated by dipolar interactions) of interaction along the c-axis. The crystallography for the present systems dictates that Jc~b >> Jc, and hence the systems should display 2D-like behavior. By this we mean that for k T >>'Jab there will be no significant correlations in the system, and the magnetic (diffuse) scattering will be uniformly spread throughout the entire Brillouin zone. As k T becomes comparable to Jab, strong correlations develop within the planes, while there will be no significant correlations between layers as has been found in the ErBa2Cu307 and DyBa2Cu307 systems just discussed. This will give rise to rods of critical scattering. For J~b > k T > J~. we will continue to see a rod of scattering; above the ordering temperature this will be critical scattering, while below the ordering temperature this will consist of low-energy spin waves. For systems which are strictly two-dimensional, of course, only an Ising model (which we believe to be appropriate for the present materials) will exhibit true long-range order at finite temperature. However, even for the cases of xy or Heisenberg spins, 3D longrange order will be induced for J~. << Jab. In the conventional two-dimensional systems such as K2NiF4 (Birgeneau et al. 1970) and K2CoF4 (Ikeda and Hirakawa 1974), and of course the Cu spins in the insulating cuprates, the 2D ordering and the 3D ordering in fact occur at (essentially) the same temperature. The reason for the induced 3D order can be understood by the following argument. Assume that a static moment has developed in the layers. There is then an energy ±JcA between the layers, where A is the average size of a domain in the layer. The minus sign is for layers that are properly matched (e.g. antiferromagnet configuration if J~: < 0), where every spin is correctly paired with the adjacent layer. The + sign, on the other hand, is for layers that are mismatched,

LANTHANIDEMAGNETICORDER: NEUTRONSCATTERINGSTUDIES

337

in which every spin is incorrectly aligned. Thus even if the interlayer coupling is very weak, there is an energy difference ~2JcA between the "correct" and the "wrong" spin configurations, and this energy difference can be quite large since A is large. Hence as soon as 2D order is established, ordering is expected to be induced along the c-axis, producing 3D Bragg peaks. In the "conventional" 2D magnets studied to date, in fact, no difference has been discerned between TN(2D) and TN(3D). An interesting exception to this behavior will be discussed in the next section. Another manifestation of the weak interactions along the c-axis is that the materials can get "confused" about what is the correct spin structure. The confusion may be caused by the influence of twin and grain boundaries, defects, intergrowths, variations in the oxygen content, etc., which can alter the delicate balance of the dipolar energies. The ErBa2Cu307 structure, for example, corresponds to chains of moments that are aligned ferromagnetically along the b-axis, while adjacent chains are aligned antiparallel to form an overall antiferromagnetic configuration. In some samples, however, the spins along the c-axis are found to be parallel rather than antiparallel, forming ferromagnetic sheets of spins (Paul et al. 1989, Chattopadhyay et al. 1988a, 1989). The dipole energies for these two configurations are very similar, and the specific structure may depend on the metallurgical state of the sample. A similar dichotomy of structures has been observed for the Dy (Clinton et al. 1991, 1995, Goldman et al. 1987) and Gd (Mook et al. 1988, Chattopadhyay et al. 1988b) 1-2-3 compounds, and a multiplicity of magnetic structures may be a common feature of these systems. The magnetic structures for the 1-2-3 materials are summarized in fig. 12; additional details are reviewed elsewhere (Lynn 1990). Dipole-dipole interactions favor three different types of magnetic order in the a - b plane based on the direction of the magnetic moment, which is generally determined by the crystal-field anisotropy. If the moment is along the c-axis (Nd, Gd, Dy), nearestneighbor spins order antiferromagnetically along both a and b directions (figs. 12c,d). If moments are along the b-axis (Er), they are coupled antiferromagnetically along a and ferromagnetically along b (figs. 12a,b), and this coupling is reversed for moments along the a-axis (fig. 12e) (Ho; Roessli et al. 1993a). All the a-b-plane spin configurations are the same for 2-4-7, 2-4-8, and for 1-2-3 compounds with different oxygen concentrations; the only known exception to this is Yb (Roessli et al. 1992), where the magnetic moment is along b like Er, but the coupling is ferromagnetic along a and antiferromagnetic along b (fig. 120. 4.3. R2Ba4Cux016 and R2Ba4Cu7015 systems - ideal 2D magnetism The 2-4-8 and 2-4-7 systems are directly related to the 1-2-3 system, and the lanthanide sublattices are shown in fig. 9 (above). For the 2-4-8 system which we will discuss first, the basic difference with the 1-2-3 is that each layer along the c-axis is shifted by b/2, producing a face-centered type of lattice along the b-axis. The nearest-neighbor distance along the c-axis is still three times a or b, and hence the intrinsic magnetic interactions are highly anisotropic just like the 1-2-3 case. The low-temperature magnetic diffraction pattern for a powder sample of Dy2Ba4CusO16 (often called simply RBa2Cu40~) is

338

J.W. LYNN and S. SKANTHAKUMAR

DyBa2Cu408 I200 3~ CO 09 LO 4d

'

'

i

.

,

.

r

i

J

,

,

,

t

,

,

,

i

'

'

'

i

.

,

,

t

,

'

'

I

,

i

i

i

iO00 800

DH © I

6OO

C),~

400

4J LO O3©

200

z~

'

0

~-~ -200

-400

,

,

,

2t

T

i

23

I

25

,

,

,

r

,

,

27

,

r

29

,

,

,

r

31

,

,

,

33

I

,

35

Scat£ering Angle 20 Fig. 13. Magnetic scattering for DyBa2Cu408 at 0.05 K, well below the ordering temperature of 0.9 K. The data are obtained by subtracting the high temperature data from the data well below the ordering temperature, and the "negative" background is due to the absence of paramagnetic scattering in the ordered state. The magnetic scattering exhibits the classic shape for a two-dimensionally ordered material. No three-dimensional Bragg peaks are observed at any temperature (Zhang et al. 1990). RBa2Cu408

O?

(a) Er

(b) Dy

I I I

O? a

(c) Ho

(d) Pr

Fig. 14. Magnetic structures for the RBa2Co408 materials. (a) Er (Zhang et al. 1990). (b) Dy (Zhang et al. 1990, Roessli et al. 1993b). (c) Ho (Roessli et al. 1994). (d) Pr (Lin et al. 1998). showa~ in fig. 13 ( Z h a n g et al. 1990). Instead o f 3D powder Bragg peaks, the scattering shows a 2D-like "sawtooth" profile even though we are well below the N r e l temperature T N - 0.9 K. The solid curve is in fact a fit to the 2D theory, a s s u m i n g long-range order within the a - b plane and no correlations along the c-direction. The scattering corresponds to the (2,g) 1 1 2D peak, which m e a n s that nearest neighbors w i t h i n the a - b plane are antiparallel. The m a g n e t i c structure for D y 2-4-8 ( Z h a n g et al. 1990, Roessli et al. 1993b) is shown in fig. 14b, along w i t h the other 2-4-8 systems investigated. In the first layer all

LANTHANIDEMAGNETICORDER: NEUTRONSCATTERINGSTUDIES

339

the spins are antiferromagnetically arranged. In the next layer up along the c-axis, all the spins are displaced by b/2. For the closest spins between layers, we will have two + spins and two -spins, and the net interaction is zero. Next-neighbor interactions also cancel, and in fact all the (point) interactions sum to zero by symmetry. Hence this is an example of a fully frustrated spin system for the interlayer coupling (indicated by the question mark in the figure), and the a-b layers are for practical purposes completely decoupled from one another. Similar behavior has been observed in the Dy2Ba4Cu7015 system (Zhang et al. 1992). In this bilayer case the magnetic scattering appears as a modulated sawtooth profile, with the modulation arising from the bilayer structure of the material. We expect a similar decoupling of adjacent layers or bilayers for any of the other lanthanide elements which have the same in-plane magnetic structure, which is in fact the most common occurrence. For Er2Ba4CusO~6, on the other hand, the Er spins in the a-b plane form ferromagnetic chains (Zhang et al. 1990), as in the Er 1-2-3 case. This magnetic structure does not possess the frustration of the Dy material, and thus conventional threedimensional Bragg peaks are observed below TN. However, above TN 2D correlations are still found, indicating that the basic interactions are 2D-like as expected. Three-dimensional magnetic Bragg peaks are also observed in Er2Ba4CuTO14.92 (B6ttger et al. 1997), with a finite but large correlation length (~130A) along the c-axis. Specific-heat data on this compound can be described by an anisotropic 2D Ising antiferromagnet, and therefore the basic interactions are 2D in character like Er l-2-3 and Er 2-4-8. The only other compound in this series that has been studied by neutron diffraction is Ho 2-4-8 (Roessli et al. 1994). The data are similar to Dy 2-4-8, and thus Ho 2-4-8 is fully frustrated for the interlayer coupling; the data show long-range 2D magnetic peaks with a short correlation length (~18 A) along the c-axis. 4.4. Oxygen dependence of the lanthanide ordering in RBa2Cu306+x Significant variations of the magnetic ordering are observed as a function of oxygen concentration for various lanthanides, such as Er (Clinton et al. 1995, Maletta et al. 1990), Dy, and Nd (Clinton et al. 1993, 1995), while in GdBa2Cu306+x the oxygen concentration does not appear to affect the nature of the three-dimensional long-range magnetic order or the ordering temperature (~2.2K) (Chattopadhyay et al. 1988b, Paul et al. 1988, Mook et al. 1988). In ErBa2Cu306+x, for example, TN is initially found to decrease with decreasing x, and fbr x ~> 0.5 long-range 3D magnetic order occurs below TN and 2D short-range correlations above TN. For x ~< 0.5 3D long-range magnetic order is lost, and only 2D correlations are observed. In ErBa2Cu306, neither 2D nor 3D long-range order develops, and only 2D short-range correlations are observed. The superconducting phase of DyBa2Cu3 06 +x exhibits the same behavior as Er, that is, no 3D correlations are observed above TN (~0.9 K), while 3D long-range order is observed below TN. In the insulating phase (x < 0.4), on the other hand, both 2D and 3D correlations develop at low temperatures, but 3D long-range order never occurs. The concentration of oxygen appears to have the strongest effects in NdBazCu306+~ (Yang et al. 1989, Clinton et al. 1993, 1995). For the fully oxygenated (and supercon-

340

J.W. LYNNand S. SKANTHAKUMAR

ducting) materials 3D long-range order is observed, with TN = 0.53 K. Modest reduction of the oxygen concentration to x = 0.78, however, completely destroys 3D order, and only short-range 2D correlations (~20 A) are observed. At lower x where the insulating phase occurs (and the Cu spins are ordered) 3D order again develops with a TN as high as 1.5 K. The transition is still sharp, as there is not much coupling to the Cu sublattice. The ordered moment for Nd is only about 1#B, so the high ordering temperature for Nd cannot be explained by dipole interactions alone. All these observations indicate that there is some coupling between lanthanide moments and the Cu chain layer, since the magnetic ordering is strongly affected by the oxygen content. The detailed nature of these interactions, however, is not understood at this time. 4.5. Pb2Sr2RI xCa~.Cu308 Neutron studies (Wu et al. 1994, 1996, Hsieh et al. 1994b, Staub et al. 1995-1997) have been performed on the magnetic ordering in powder samples of Pb2Sr2R1 ~CaxCu3Os, which superconducts for 0.2 < x <0.8. In these materials all the lanthanides except Ce superconduct, with Tc as high as 80K. The suppression of superconductivity in Pb2Sr2Cel-xCaxCu308 is not thought to be connected to the magnetism, but to the tetravalent nature of the Ce ion in this compound (Skanthakumar and Soderholm 1996). The lanthanide ions in this system have a site environment that is similar to that in the 1-2-3 systems, and the 2D magnetic behavior that has been observed for R = Tb and Pr is not unexpected. The separation along the c-axis is about 15.8 A, which is even larger than the 1-2-3 system. In Pb2Sr2Tbl-xCaxCu308, the Tb spins are found to order with relatively high transition temperatures of 5.3 and 4 K for x = 0.0 and 0.5, respectively. These high ordering temperatures again demonstrate that exchange interactions dominate in these materials. In addition, magnetic susceptibility and inelastic neutron scattering experiments suggest that significant 2D T b - T b correlations persist even at temperatures above the superconducting transition temperature, which is more than an order of magnitude above TN. Further work on this class of materials is warranted.

5. The special case of Pr A singular exception to the low ordering temperatures of the lanthanides is found for the class of cuprates that contain Pr: these generally are not even metallic let alone superconducting. This unique behavior has inspired extensive studies of the physical properties of the Pr cuprates in order to determine why only Pr suppresses the superconducting state while the other lanthanides exhibit very little effect. The magnetic ordering of the Pr is also anomalous in that the ordering temperatures are an order of magnitude higher (~ 15 K) than for the other rare-earth systems, indicating much stronger exchange and hybridization effects for the 4f Pr electrons. Investigations of the magnetic order of Pr in PrBa2Cu306+x have been complicated by a number of factors. One is that this material is a semiconductor for all x, and thus

LANTHANIDEMAGNETICORDER: NEUTRON SCATTERINGSTUDIES

341

the Cu spins always order at a much higher temperature than the Pr spins. A second complication is that with appropriate doping, either directly on the chain site or on the Ba site, the chain spins can also develop a moment and exhibit long-range magnetic order, either together with the plane spins or as a separate magnetic transition. Moreover, the development of chain moments turns out to be especially sensitive to even small levels of dopants in the Pr 1-2-3 system, which has been a particular problem with many fluxgrown crystals. A further complication is that the ordered moments for Pr and Cu are comparable in magnitude, and this can lead to ambiguities in the interpretation of data. Finally, Pr has a tendency to occupy the Ba site in Pr 1-2-3, and this changes both the magnetic and electronic behavior. These complications have led to possible ambiguities in interpretations and lively debates, but a consensus has now emerged about the nature of the Pr ordering in these materials. Nevertheless, the consensus is certainly not yet unanimous, and the debate will no doubt continue with the recent report by Zou et al. (1998) of superconductivity in a small fraction of a variant, not yet identified, phase of Pr 1-2-3. This contrasts with the semiconducting behavior found in the overwhelming majority of Pr 1-2-3 samples, as already noted. However, it has been shown that the susceptibility in the sample that exhibits superconductivity is a factor of two smaller than was reported initially (Narozhnyi and Drechsler 1999, Zou et al. 1999), suggesting that the Pr content is reduced by half, and thus it seems likely that the floating-zone technique produces small regions that are either Y-rich and/or Ba-rich with Ba occupying the Pr site. In particular, it is known that relatively modest substitution for Ca on the Pr site provides a 90 K superconductor (Xiong et al. 1998), and similar behavior would be expected for Ba. Thus it seems likely that the superconductivity does not originate from Pr 1-2-3, but rather from an alloy. We start our discussion of the Pr magnetic ordering by considering PrBa2Cu306+x, which was the first Pr material to be investigated with neutrons (Li et al. 1989). The system is a semiconductor for the full range of x, with the Cu ions retaining their high magnetic ordering temperature. The f-electron hybridization has been observed directly by inelastic neutron scattering experiments, which have found that there are substantial linewidths to the crystal-field excitations of the Pr (Staub and Soderholm 2000). The exchange interactions are thereby enhanced, increasing the ordering temperature and rendering the lanthanide ordering 3D in nature. In this initial study a simple antiferromagnetic ordering was found to develop below TN = 17K, with antiferromagnetic coupling between nearest-neighbor Pr moments in all three directions in the crystal. The magnetic structure is the same as found in the Dy and Gd 1-2-3 systems, shown in fig. 12c. The observed ordering temperature was in good agreement with specific heat and susceptibility results, which studies have elucidated the systematics of this ordering by following it as a function of oxygen concentration and as a function of Y (and other lanthanides) substitution on the Pr site (for a review see Radousky 1992). Bulk measurements show that as oxygen is removed TN(Pr) monotonically decreases to 12 K for PrBa2Cu306, and this has been confirmed by neutron diffraction measurements on the depleted system (Guillaume et al. 1993).

342

J.W. LYNN and S. S K A N T H A K U M A R

Studies on single crystals of the "pure" PrBa2Cu306+x have been carried out subsequently, but the results have been complicated because of contamination of the samples from the crucible, and/or Pr substitution on the Ba site. In particular, it is now known that crystals grown in alumina or MgO crucibles by the flux technique become doped by A1 or Mg (Casalta et al. 1994, Uma et al. 1998b). These (inadvertent) dopings can cause the Cu chain spins to order (Lynn et al. 1988, Li et al. 1988, 1990, Rosov et al. 1992a). This ordering varies with oxygen content x, and is strongly coupled to the Pr sublattice because the symmetry of the two magnetic structures is the same. The chain doping can also reduce the N6el temperature for the Pr ordering, and can change the nature of the Pr structure along the c-direction (Longmore et al. 1996, Boothroyd et al. 1997, Boothroyd 1998, 2000) so that the usual c-axis antiferromagnetic coupling is not found in these samples. During the course of these investigations there was also a suggestion from NMR measurements (Nehrke and Pieper 1996) that the Pr was actually non-magnetic in pure Pr 1-2-3, but a reinvestigation with neutrons (Skanthakumar et al. 1997b) on pure powders indicated that the Pr was indeed ordering, and this has also been confirmed on pure crystals by neutrons (Uma et al. 1998a) and by resonant X-ray scattering (Hill et al. 1998). It therefore appears likely that the NMR may have been detecting some Pr on the Ba site in that sample, which may well be in a nonmagnetic singlet crystal-field state, while the line due to the Pr on the R site is too broad to observe. The relatively recent development of non-reactive BaZrO3 crucibles has now permitted the growth and investigation of undoped single crystals. In a fully oxygenated single crystal the Cu spins order at 281 K, while the Pr moments order at 16.8K as shown in fig. 15 (Uma et al. 1998a). The initial ordering is revealed by the intensity of the I 1 (~,g,0) versus T shown in the center. This structure corresponds to antiferromagnetic alignment of spins in the a-b plane, while the spins along the c-direction are ferromagnetically aligned. The development of the ordered Pr moment is accompanied by a modest coupling to the Cu spins as indicated by the intensity of the (g,g,2) ~ 1 Cu peak (top). At ~ 11 K (on cooling) a first-order "spin-flop" transition occurs to a Pr spin structure where the nearest-neighbor Pr moments along the c-axis flip to become antiparallel rather than parallel. This corresponds to the development of the (g,g,g)-type 1 1 1 peak as indicated in the bottom of the figure. The data clearly indicate that there is strong hysteresis in both types of Pr magnetic peaks, and the behavior is in good agreement with the specific heat and thermal expansion data taken on warming and cooling. The results on this pure single crystal are in general agreement with the data obtained on the polycrystalline sample, but the details with regard to the crossover from ferromagnetic to antiferromagnetic nearest-neighbor alignment along the c-axis are different. These differences are likely due to the delicate balance of interactions along the c-axis (Narozhnyi et al. 1999), and their subtle dependence on factors such as small concentrations of defects and impurities, oxygen content, strain, etc. There have also been indications of a small in-plane incommensurability of the magnetic peaks observed by X-ray resonant scattering (Hill et al. 1998). Studies have also been carried out to observe the effects on both Cu and Pr order by chemical substitution on other sites in the Pr 1-2-3 system. Zn is found to substitute on

LANTHANIDE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES

343

PrBazCu307

c_ E

o Q)

,ooF zoo

.

( I. / 2. . .1 /. Z. , 0. ) .

oI~ i s ~^

.

.

.

^

, ~ , ,

F~[

" •. . . warming . . . . ,~ coo~i.g

.

. , e~,~

r;

%

5oo ' , 1 / 2 , 1 / 2 , 1 / 2 )

_

J

~

1000

.

|

.......

©

cooling

l

I

;

? '

....

I .... 5

I ,q~ 10

, ~ 15

Temperature (K)

20

25

Fig. 15. Temperature dependence of the (top) (½,½,2), (center) (½,½,0), (bottom) (~,~,~) Ill magnetic Bragg peaks in a pure single crystal of PrBa2Cu307 (Urea et al. 1998a).

the Cu planes, and this has no effect on either the Pr ordering temperature or size of the ordered moment (Li et al. 1993), but it does change the coupling along the c-axis from antiferromagnetic to ferromagnetic. Ga, on the other hand, substitutes preferentially on the Cu chain sites. This again changes the magnetic structure along the c-axis from antiferromagnetic to ferromagnetic, but in addition TN is reduced (Li et al. 1994) while the ordered moment remains unchanged. Neither of these substitutions has a significant effect on the Cu powder Bragg peaks in the temperature regime where the Pr orders. Another type of substitution can occur for "pure" PrBa2Cu306+x, where the Pr can substitute on the Ba site forming Pr~ +xBa2--xCu306+y. This has the effect of reducing the ordering temperature for the Pr (Malik et al. 1992), as does La substitution for Ba (Wang et al. 2000). There has been considerable additional work investigating the magnetic ordering of the Pr in the cuprate class of materials, and the properties for Pr systems investigated with neutrons in this class of materials have been reviewed recently elsewhere (Lynn 1997). PrBa2Cu408 also orders antiferromagnetically at TN ~ 17 K like the 1-2-3 material, with an analogous magnetic structure (Lin et al. 1998, Li et al. 1999). For PrBa2Cu2NbO8 (Rosov et al. 1993) the Cu chain layer is replaced with a (fully oxygenated) NbO2 layer, which of course carries no moment. This eliminates the complication of the chain magnetism, and we still find a high ordering temperature for the Pr and no influence on the Cu plane ordering when the Pr subsystem orders. A similar situation occurs for T1Ba2PI"Cu207 (Hsieh et al. 1994a) where the CuO chains are replaced by T10.

344

J.W. LYNNand S. SKANTHAKUMAR

There is no observable interaction between the Cu and Pr order, as would be expected since the magnetic structures have different symmetries. Similar behavior is observed for the more complicated Prl.sCe05Sr2Cu2(Nb,Ta)Ol0 system (Goodwin et al. 1997). Again there can be no Cu chain magnetism, and there is a high observed TN for the Pr. The Pb2Sr2PrCu308 material (Hsieh et al. 1994b, Lin et al. 1997) is somewhat different in that it has no oxygen in the Cu chain layer, while the properties of the Pr are quite similar to the other systems. Finally, we note that the related BaPrO3 material, which obviously has no complications of Cu ordering of any kind, exhibits a similar Pr ordering temperature and reduced moment (Rosov et al. 1992b) as for the cuprates. Thus in all the layered cuprates the Pr carries a moment and orders magnetically at much higher temperatures than the heavy lanthanides.

6. Lanthanide spin dynamics Measurements of the spin dynamics of the lanthanide moments are considerably more difficult than determining the magnetic structures, but several investigations have been made which we briefly point out. If we introduce lanthanide exchange interactions, then this will give rise to magnetic excitons propagating in the lattice. This is revealed as dispersion of the crystal-field levels, and such dispersion has been observed for both Pr2CuO4 (Sumarlin et al. 1994, 1995) and Nd2CuO4 (Henggeler et al. 1996), and for the Ce-doped systems (Henggeler et al. 1997). The observed dispersion is as large as a few meV, which indicates that there are quite significant R - R exchange interactions in these materials. The spin dynamics of the exchange-split crystal-field ground state have also been investigated for Nd2CuO4 by Loewenhaupt et al. (1995), and more recently by Casalta et al. (1998) above the Nd magnetic ordering temperature. The latter authors find that there is a fluctuation component that persists up to room temperature, and is associated with the exchange field of the Cu spins acting on the Nd moments. This has decidedly 2D character and presumably originates from the 2D character of the Cu spin system. The other component originates from the usual paramagnetic spin fluctuations of the Nd moments above the ordering temperature. Below TN the Nd and Cu modes are predicted to be mixed in the parent compound, and the theoretical predictions for the spin dynamics in the ordered state (Thalmeier 1996, Sachidanandam et al. 1997) are in reasonable agreement with experiment (Henggeler et al. 1997). There is predicted to be an additional mode that has yet to be observed, and additional work in this area is desirable. In the doped system the observed spin waves are simplified because the Cu spins are not ordered, and the model calculations are able to reproduce the data. The spin dynamics of the lanthanide systems have also been investigated both above and below the ordering temperature in superconducting ErBa2Cu307 (Skanthakumar et al. 1997a). In this case the Cu moments interact weakly with the Er and are not ordered, and the lanthanide system can thus be treated independently of the Cu. The groundstate dispersion relation along the [110] direction is shown in fig. 16. There is a large (compared to kTy) gap of about 0.20meV in the excitation spectrum, as would be

LANTHANIDE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES

ErBa Cu 0 2

~

l

'

'

'

I

i ,

I T ,

i

i

I ' '

8

I

I

I

I

345

7 I

I

I

'

'

I

I

I

0.24

> o E 0.16 V

(h,h, 7.s)

o r-. 0 . 0 8

T=O.4K

1.1.1

-g.

0'4- .

'

o ......

h

0.'4

" 0.s

Fig. 16. Spin wave dispersion along [110] direction for ErBa2Cu307. The solid curve is a guide to the eye (Skanthakumar et al. 1997a).

expected for an Ising system. However, for a pure Ising system S z commutes with the Hamiltonian and thus the spin waves would be dispersionless, while there is clearly some observed dispersion in the a-b plane. Measurements along the c-axis indicate that these excitations have no measurable dispersion in that direction, and this behavior is expected because the magnetic interactions are weak along the c-direction. The Ising-type gap is still dominant, though, and the data of fig. 16 show that this material is still a good approximation to an Ising magnet. Finally, we note that the spin dynamics for the Pr 1-2-3 system have been investigated in some detail, and not surprisingly, it is quite different than the other systems. A broad distribution of quasielastic scattering (and broad crystal field levels) was observed in powders (Skanthakumar et al. 1990). These energy widths could be due to exchange interactions, where the width is caused by the powder average of a dispersive excitation, or it could be intrinsically broad due to hybridization. Single-crystal measurements reveal a broad, dispersionless distribution of scattering (Lynn et al. 2000), indicating that hybridization is the cause, in addition, the distribution of magnetic scattering is temperature-independent, similar to what has been observed in non-Fermi liquid systems (Aronson et al. 1995). Lister et al. (2000) have also carried out inelastic measurements on single crystals recently and find a similar distribution of scattering, but with some structure in it. Further measurements will be needed to explore these details.

7. O v e r v i e w and future directions

The magnetic ordering of the lanthanide ions in superconducting systems has been a topic of active interest for many years. In conventional "magnetic superconductors" such as the Chevrel phase and related materials the lanthanide moments are coupled very weakly both to the metallic electrons and to each other, resulting in very low

346

J.W. LYNNand S. SKANTHAKUMAR

magnetic ordering temperatures and a delicate energetic balance with superconductivity. A similar situation occurs for the layered cuprates in that the development of longrange lanthanide magnetic order also occurs at very low temperatures. In contrast to the earlier systems, though, the superconductivity in the cuprates typically occurs at much higher temperatures, and the antiferromagnetic order that develops on the lanthanide sublattice coexists with the superconducting state. Neutron scattering has revealed the nature of the magnetic ordering of the lanthanide ions in the layered 1-2-3, 2-4-8, and 2-4-7 oxide superconductors, where the separation of the lanthanide ions is much larger along the c-axis than along the a-b directions. This renders these materials prototypical two-dimensional (2D) magnets, and with the spin-spin interactions being relatively weak compared to the anisotropies, this generally results in Ising-like magnetic behavior. Thus in the ErBa2Cu307 material, for example, there is remarkable agreement between the observed order parameter and the exact solution for the S = ½ 2D Ising model. Another textbook example is provided by the DyBa2Cu408 system, where a geometric cancellation of the already weak interactions occurs along the c-axis, effectively decoupling the lanthanide a-b layers. The system thus exhibits no crossover to the 3D behavior usually found below the ordering temperature, making it the best example of a 2D magnet known to date. One of the interesting avenues of exploration will be to investigate the spin dynamics of these materials, as high-quality single crystals large enough for inelastic scattering become available. The 2-1-4 class of materials, on the other hand, provides the first examples of superconductors where the exchange interactions unambiguously dominate the lanthanide energetics, such as Sm2CuO4 which has a dipole ordering temperature that is two orders of magnitude lower than the observed ordering temperature of 6 K. Thus the traditional picture of negligible exchange being necessary for a superconducting material to simultaneously exhibit long-range magnetic order has to be abandoned. Of course, the magnetic properties of the Cu ions are also of central interest, where the magnetic fluctuations may be directly involved in the formation of the superconducting state. The question of the origin of the superconductivity has by no means been settled yet, but even if magnetic fluctuations are not at the root of the superconductivity in these high-To materials, it is quite clear that the Cu-O layers are intimately involved in both the magnetism and superconductivity, and the striking magnetic behavior these materials display is of fundamental interest in its own right. From the very early days of cuprate superconductors it was known that a dramatic exception to the overall behavior of these magnetic-superconductor systems occurs for Pr, and this singular exception has been the subject of extensive research. The magnetic ordering temperatures are an order of magnitude higher for Pr than for the other lanthanide materials, and consequently the magnetic coupling must be dominated by exchange. Moreover, with the exception of PrzCuO4 where the crystal-field ground state is an isolated singlet, superconductivity is significantly affected in the Pr materials. The origins of this effect and the nature of the Pr magnetism are still under active investigation and debate, and this will certainly be an area that will receive attention from experimentalists and theorists alike.

LANTHANIDE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES

347

Acknowledgments We w o u l d like to thank all o f our collaborators w h o have w o r k e d w i t h us on various aspects o f these problems. We thank Taner Y i l d i r i m for providing the calculations s h o w n in figs. 5 and 7. We w o u l d also like to thank Tarter, Frederic Bourdarot, and D a n Derider for a careful reading o f the manuscript. W o r k at A r g o n n e is supported by the D O E B E S , C h e m i c a l Sciences under contract W - 3 1 - 1 0 9 - E N G - 3 8 . Work at the U n i v e r s i t y o f M a r y l a n d supported in part by the NSF, D M R 97-01339.

References Abrikosov, A.A., and L.E Gorkov, 1961, Sov. Phys. JETP 12, 1243. Akimitsu, J., H. Sawa, T. Kobayashi, H. Fujiki and Y. Yamada, 1989, J. Plays. Soc. Jpn. 58, 2646. Allenspach, R, S.-W. Cheong, A. Dommann, R Fischer, Z. Fisk, A. Furrer, H.R. Ott and B. Rupp, 1989, Z. Phys. B 77, 185. Andreev, A.V., A.I. Buzdin and R.M. Osgood, 1990, JETP Lett. 52, 55. Aronson, M.C., R. Osborn, R.A. Robinson, J.W. Lynn, R. Chau, C.L. Seaman and M.B. Maple, 1995, Phys. Rev. Lett. 75, 725. Bacon, G.E., 1975, Neutron Diffraction, 3rd Ed. (Oxford University Press, Oxford). Birgeneau, R.J., H.J. Guggenheim and G. Shirane, 1970, Phys. Rev. B 1, 2211. Boothroyd, A.T., 1998, Physica B 241-243, 792. Boothroyd, A.T., 2000, J. Alloys & Compounds 303-304, 489. Boothl"oyd, A.T., A. Longmore, N.H. Andersen, E. Brecht and Th. Wolf, 1997, Phys. Rev. Lett. 78, 130. B6ttger, G., R Fisher, A. DSnni, P. Berastegui, Y. Aoki, H. Sato and E Fauth, 1997, Phys. Rev. B 55, 12005. Casalta, H,, R Schleger, E. Brecht, W. Montfi'ooij, N.H. Andersen, B. Lebech, W.W. Schmahl, H. Fuess, R. Liang, W.N. Hardy and T. Wolf, 1994, Phys. Rev. B 50, 9688. Casalta, H., Ph. Bourges, M. D'Astuto, D. Petitgrand and A.S. Ivanov, 1998, Phys. Rev. B 57, 471. Chattopadhyay, T., and K. Siemensmeyer, 1995, Europhys. Lett. 29, 579, Chattopadhyay, T., EL Brown, D. Bonnenberg, S. Ewert and H. Maletta, 1988a, Europhys. Lett. 6, 363. Chat~opadhyay, T., H. Maletta, W. Wirges, K. Fisher and RJ. Brown, 1988b, Phys. Rev. B 38, 838.

Chattopadhyay, T., EJ. Brown, B.C. Sales, L.A. Boatnet, H.A. Mook and H. Maletta, 1989, Phys. Rev. B 40, 2624. Chattopadhyay, "12, O.B. HyuI1 and D.K. Finnemore, 1990, Solid State Commun. 73, 721. Chattopadhyay, T., EJ. Brown and U. Kobler, 1991a, Physica C 177, 294. Chattopadhyay, T., P.J. Brown, A.A. Stepanov, R Wyder, J. Voiron, A.I. Zvyagin, S.N. Bafilo, D.I. Zhigunov and I.A. Zobkalo, 1991b, Phys. Rev. B 44, 9486. Chattopadhyay, T., RJ. Brown, B. Roessli, A.A. Stepanov, S.N. Barilo aud D.I. Zhigunov, 1992, Phys. Rev. B 46, 5731. Chattopadhyay, T., EL Brown and B. Roessli, 1994a, J. Appl. Phys. 75, 6816. Chattopadhyay, T., J.W. Lynn, N. Rosov, T.E. Grigereit, S.N. Barilo and D.I. Zhigunov, 1994b, Phys. Rcv. B 49, 9944. Clinton, T.W., J.V~LLynn, J.Z. Liu, Y.X. Jia and R.N. Shelton, 1991, J. Appl. Phys. 70, 5751. Clinton, T.W., J.W. Lynn, B.W. Lee, M. Buchgeister and M.B. Maple, 1993, J. Appl. Phys. 73, 6320. Clinton, TW., J.W. Lynn, J.Z. Liu, Y.X. Jia, T.J. Goodwin, R.N. Shelton, B.W. Lee, M. Buchgeister, M.B. Maple and J.L. Peng, 1995, Phys. Rev. B 51, 15429. Cox, D.E., A.I. Goldman, M.A. Subramanian, J. Gopalakrishnan and A.W. Sleight, 1989, Plays. Rev. B 40, 6998. Elschner, B., and A. Loidl, 2000, in: Handbook on the Physics and Chemistry of Rare Earths, Vol. 30, eds K.A. Gschneidner Jr, L. Eyring and M.B. Maple (Elsevier, Amsterdam) ch. 191. Endoh, Y., M. Matsuda, K. Yamada, K. Kakurai, Y. Hidaka, G. Shirane and R.J. Birgeneau, 1989, Phys. Rev. B 40, 7023.

348

J.W. LYNN and S. SKANTHAKUMAR

Fischer, O., and M.B. Maple, eds, 1983, Superconductivity in Ternary Compounds, Vols. 32 and 34 of Topics in Current Physics (Springer, New York). Fischer, R, K. Kalau-ai, M. Steiner, K.N. Clanscn, B. Lebech, E Hulliger, H.R. Ott, R Briiesch and R Unternfihrer, t988, Physica C 152, 145. Fischer, E, B. Schmid, R Briiesch, E Stucki and R Untern/ihrer, 1989, Z. Phys. B 74, 183. Goldman, A.I., B.X. Yang, J.M. Tranquada, J.E. Crow and C.-S. Jee, 1987, Phys. Rev. B 36, 7234. Goodwin, T.J., R.N. Shelton, H.B. Radousky, N. Rosov and J.W. Lynn, 1997, Phys. Rev. B 55, 3297. Guillaume, M., R Fischer, B. Roessli, A. Podlesnyak, J. Schefer and A. Fnrrer, 1993, Solid State Commnn. 88, 57. Guillaume, M., R Fischer, B. Roessli, P. Allenspach and V Tronnov, 1994, Physica C 235, 1637. Gukasov, A.G., S.Yu. Kokovin, VP. Plakhty, I.A. Zobkalo, S.N. Baillo and D.I. Zhigtmov, 1992, Physica B 180, 455. Henggeler, W., T. Chattopadhyay, B. Roessli, D.I. Zhignnov, S.N. Barilo and A. Furrer, 1996, Z. Plays. B 99, 465. Henggeler, W., T. Chattopadhyay, B. Roessli, P. Vorderwisch, P. Thalmeier, D.I. Zhignnov, S.N. Baillo and A. Furrer, 1997, Phys. Rev. B 55, 1269. Hill, J.R, A. Vigliante, D. Gibbs, J.L. Peng and R.L. Greene, 1995, Phys. Rev. B 52, 6575. Hill, J.R, A.T. Boothroyd, N.H. Andersen, E. Brecht and Th. Wolf, 1998, Phys. Rev. B 58, 11211. Hsieh, W.T., K.J. Chang, W.-H. Li, K.C. Lee, J.W. Lynn, C.C. Lai and H.C. Ku, 1994a, Phys. Rev. B 49, 12200. Hsieh, W.T., W-H. Li, K.C. Lee, J.W. Lynn, J.H. Shieh and H.C. Ku, 1994b, J. Appl. Phys. 76, 7124. lkeda, H., and K. Hirakawa, 1974, Solid State Commun. 14, 529. Katano, S., R.M. Nicklow, S. Funahashi, N. Moil, T. Kobayashi and J. Akimitsu, 1993, Physica C 215, 92. Li, W.-H., J.W. Lynn, H.A. Mook, B.C. Sales and Z. Fisk, 1988, Phys. Rev. B 37, R9844. Li, W-H., J.W. Lynn, S. Skanthakumar, T.W. Clinton, A. Kebede, C.-S. Jee, J.E. Crow and T. Mihalisin, 1989, Phys. Rev. B 40, R5300. Li, W.-H., J.W. Lynn and Z. Fisk, 1990, Phys. Rev. B 41, 4098. Li, W.-H., K.J. Chang, W.T. Hsieh, K.C. Lee, J.W. Lynn and H.D. Yang, 1993, Phys. Rev. B 48, 519.

Li, W.-H., C.J. Jou, S.T. Shyr, K.C. Lee, J.W. Lynn, H.L. Tsay and H.D. Yang, 1994, J. Appl. Phys. 76, 7136. Li, W.-H., W.Y. Chuang, S.Y. Wu, K.C. Lee, J.W. Lynn, H.L. Tsay and H.D. Yang, 1997, Phys. Rev. B 56, 5631. Li, W.-H., S.Y. Wu, Y.-C. Lin, K.C. Lee, J.W. Lynn, S.S. Weng, I.E Hong, J.-Y. Lin and H.D. Yang, 1999, Phys. Rev. B 60, 4212. Lin, Y.-C.,. W.Y. Chuang, W.-H. Li, K.C. Lee, J.W. Lynn, C.L. Yang and H.C. Ku, 1997, J. Appl. Phys. 81, 4940. Lin, Y.-C., S.Y. Wu, W.-H. Li, K.C. Lee, J.V~ Lynn, C.W. Lin, J.-Y. Lin and H.D. Yang, 1998, Physica B 241-243, 702. Lister, S.J.S., A.T. Boothroyd, N.H. Andersen, A.A. Zhokhov and A.N. Christensen, 2000, Physica B 276, 799. Loewenhaupt, M., R Fabi, S. Horn, R v. Aken and A. Severing, 1995, J. Magn. Magn. Mater. 140-144, 1293. Longmore, A., A.T. Boothroyd, C. Changkang, H. Yongle, M.E Nutley, N.H. Andersen, H. Casalta, P. Schleger and A.N. Christensen, 1996, Phys. Rev. B 53, 9382. Lynn, J.W., ed., 1990, High Temperature Superconductivity (Springer, New York). Lynn, J.W., 1992, J. Alloys & Compounds 181,419. Lynn, J.W., 1997, J. Alloys & Compounds 250, 552. Lynn, LW., J.A. Gotaas, R.W. Erwin, R.A. Ferrell, J.K. Bhattachaljee, R.N. Shelton and R Klavins, 1984, Phys. Rev. Letters 52, 133. Lynn, J.W., W.-H. Li, Q. Li, H.C. Ku, H.D. Yang and R.N. Shelton, 1987, Phys. Rev. B 36, R2374. Lynn, J.W., W.-H. Li, H.A. Mook, B.C. Sales and Z. Fisk, 1988, Phys. Rev. Lett. 60, 2781. Lynn, J.W., T.W. Clinton, W.-H. Li, R.W. Erwin, J.Z. Liu, K. Vandervoort, R.N. Shelton and E Klavins, 1989, Phys. Rev. Lett. 63, 2606. Lynn, J.W., I.W. Sumarlin, S. Skanthakumar, W.-H. Li, R.N. Shelton, J.L. Peng, Z. Fisk and S.-W. Cheong, 1990, Phys. Rev. B 41, R2569. Lynn, J.W., S. Skanthakumar, Q. Huang, S.K. Sinha, Z. Hossain, L.C. Gupta, R. Nagarajan and C. Godart, 1997, Phys. Rev. B 55, 6584. Lynn, J.W., N. Rosov, S.N. Barito, L. Kurnevitch and A. Zhokhov, 2000, Chin. J. Phys. 38, 286. Maletta, H., E. P6rschke, T. Chattopadhyay and PJ. Brown, 1990, Physica C 166, 9.

LANTHANIDE MAGNETIC ORDER: NEUTRON SCATTERING STUDIES Malik, S.K., S.M. Pattalwar, C.V Tomy, R. Prasad, N.C. Soni and K. Adhikary, 1992, Phys. Rev. B 46, 524. Maple, M.B., 1976, Appl. Phys. 9, 179. Maple, M.B., 2000, in: Handbook on the Physics and Chemistry of Rare Earths, Vol. 30, eds K.A. Gschneidner Jr, L. Eyring and M.B. Maple (Elsevier, Amsterdam) ch. 187. Matsuda, M., K. Yamada, K. Ka~rai, H. Kadowaki, T.R. Thurston, Y. Endoh, Y. Hidaka, R.J. Birgeneau, M.A. Kastner, P.M. Gehring, A.H. Moudden and G. Shirane, 1990, Phys. Rev. B 42 10098. Matsuda, M., Y. Endoh, K. Yamada, H. Kojima, I. Tanaka, R.J. Birgeneau, M.A. Kastner and G. Shirane, 1992, Phys. Rev. B 45, 12548. Mook, H.A., D.McK. Paul, B.C. Sales, L.A. Boatner and L. Cussen, 1988, Phys. Rev. B 38, 12008. Narozhnyi, VN., and S.-L. Drechsler, 1999, Phys. Rev. Lett. 82, 461. Narozhnyi, V.N., D. Eckert, K.A. Nenkov, G. Fuchs, T.G. Uvarova and K.-H. Mtiller, 1999, Physica C 312, 233. Nehrke, K., and M.W. Pieper, 1996, Phys. Rev. Lett. 76, 1936. Onsager, L., 1944, Phys. Rev. 65, 117. Paul, D.McK., H.A. Mook, A.W. Hewit, B.C. Sales, L.A. Boatner, J.R. Thompson and M. Mostoller, 1988, Phys. Rev. B 37, 2341. Paul, D.McK., H.A. Mook, L.A. Boatner, B.C. Sales, J.O. Ramey and L. Cussen, 1989, Phys. Rev. B 39, 4291. Petitgrand, D., A.H. Moudden, E Galez and R Boutrouille, 1990, J. Less-Common Met. 164-165, 768. Petitgrand, D., L. Boudarbne, Ph. Bourges and P. Galez, 1992, J. Magn. Magn. Mater. 104 107, 585. Radaelli, P.G., J.D. Jorgensen, A.J. Schultz, J.L. Peng and R.L. Greene, 1994, Phys. Rev. B 49, 15322. Radousky, H.B., 1992, J. Mater. Res. 7, 1917. Roessli, B., P. Allenspach, P. Fisher, J. Mesot, U. Staub, H. Maletta, P. Brfiesch, C. Ritter and A.W. Hewat, 1992, Physica B 180&181, 396. Roessli, B., R Fisher, U. Staub, M. Zolliker and A. Furrer, 1993a, Europhys. Lett. 23, 511. Roessli, B., R Fisher, M. Zolliker, P. Allenspach, J. Mesot, U. Staub, A. Furrer, E. Kaldis, B. Bucher, J. Karpinski, E. Jilek and H. Mntka, 1993b, Z. Phys. B 91, 149. Roessli, B., R Fisher, M. Guitlaume, J. Mesot, U. Staub, M. Zolliker, A. Furrer, E. Kaldis, J. Karpinski

349

and E. Jilek, 1994, J. Phys.: Condens. Matter 6, 4147. Rosov, N., J.W Lynn, G. Cao, J.W. O'Reilly, R Pernambuco-Wise and J.E. Crow, 1992a, Physica C 204, 171. Rosov, N., J.W Lynn, Q. Lin, G. Cao, J.W O'Reilly, R Pernambuco-Wise and J.E. Crow, 1992b, Phys. Rev. B 45, 982. Rosov, N., J.W Lynn, H.B. Radousky, M. Bennahmias, T.J. Goodwin, R Klavins and R.N. Shelton, 1993, Phys. Rev. B 47, 15256. Rosseinsky, M.J., K. Prassides and E Day, 1989, J. Chem. Soc., Chem. Commun. 22, 1734. Rosseinsky, M.J., K. Prassides and P. Day, 1991, Inorg. Chem. 30, 2680. Sachidanandam, R., T. Yildirim, A.B. Harris, A. Aharony and O. Entin-Wohlman, 1997, Phys. Rev. B 56, 260. Skanthakumar, S., 1993, Ph.D. Thesis (University of Maryland). Skanthakumar, S., and L. Soderholm, 1996, Phys. Rev. B 53, 920. Skanthakumar, S., H. Zhang, ZW. Clinton, W-H. Li, J.W Lynn, Z. Fisk and S.-W Cheong, 1989, Physica C 160, 124. Skanthakumar, S., W-H. Li, J.W Lynn, A. Kebede, LE. Crow and T. Mihalisin, 1990, Physica B 163, 239. Skanthakumar, S., J.W. Lynn, J.L. Peng and Z.Y. Li, 1991, J. Appl. Phys. 69, 4866. Skanthakumar, S., J.W. Lynn, J.L. Peng and Z.Y. Li, 1992, J. Magn. Magn. Mater. 104-107, 519. Skanthakumar, S., J.W. Lynn, J.L. Peng and Z.Y. Li, 1993a, J. Appl. Phys. 73, 6326. Skanthakumar, S., J.W. Lynn, J.L. Peng and Z.Y~ Li, 1993b, Phys. Rev. B 47, R6173. Skanthakumar, S., J.W Lynn and E Dogan, 1997a, J. Appl. Phys. 81, 4934. Skanthakumar, S., J.W. Lynn, N. Rosov, G. Cao and J.E. Crow, 1997b, Phys. Rev. B 55, R3406. Skanthakumar, S., J.W Lynn, J.L. Peng and Z.Y. Li, 1999, private communication. Soderholm, L., S. Skanthakumar and C.W Williams, 1999, Phys. Rev. B 60, 4302. Stassis, C., and A.I. Goldman, 1997, J. Alloys & Compounds 250, 603. Staub, U., and L. Soderholm, 2000, in: Handbook on the Physics and Chemistry of Rare Earths, Vol. 30, eds K.A. Gschneidner Jr, L. Eyring and M.B. Maple (Elsevier, Amsterdam) ch. 194.

350

J.W LYNN and S. SKANTHAKUMAR

Staub, U., L. Soderholm, S. Skanthakumar and M.A. Antonio, 1995, Plays. Rev. B 52, 9736. Staub, U., L. Soderholm, S. Skanthakumar, S. Roskenkranz, C. Ritter and Vv~ Kagunya, 1996, Europhys. Lett. 34, 447. Staub, U., L. Soderholm, S. Skanthakumar, S. Roskenkranz, C. Ritter and W. Kagunya, 1997, Z. Phys. B 104, 37. Sumarlin, I.W., S. Skanthakumar, J.W. Lynn, J.L. Peng, W. Jiang, Z.Y. Li and R.L. Greene, 1992, Phys. Rev. Lett. 68, 2228. Sumarlin, I.W., J.VKLynn, T. Chattopadhyay, S.N. Barilo and D.I. Zhigunov, 1994, Physica C 219, 195. Sumarlin, I.W., J.W Lynn, T. Chattopadhyay, S.N. Barilo, D.I. Zhigunov and J.L. Peng, 1995, Phys. Rev. B 51, 5824. Thalmeier, R, 1996, Physica C 266, 89. Thomlinson, W., G. Shirane, J.W. Lynn and D.E. Moncton, 1983, in: Superconductivity in Ternary Compounds, eds M.B. Maple and O. Fischer, Vol. 34 of Topics in Current Physics (Springer, New York) ch. 8. Thurston, T.R., M. Matsuda, K. Kakurai, K. Yamada, Y. Endoh, R.J. Birgenean, EM. Gehring, Y. Hidaka, M.A. Kastner, T. Murakami and G. Shirane, 1990, Phys. Rev. Lett 65, 263. Uma, S., W. Schnelle, E. Gmelin, G. Rangarajan, S. Skanthakumar, J.W Lynn, R. Walter, T. Lorenz, B. Bucher and A. Erb, 1998a, J. Phys. (Lett.) 10, L33. Uma, S., G. Rangarajan and E. Gmelin, 1998b, Physica C 301, 141. Vaknin, D., S.K. Sinha, D.E. Moncton, D.C. Johnston, J.M. Newsam, C.R. Safinya and H.E. King Jr, 1987, Phys. Rev. Lett. 58, 2802. Vaknin, D., S.K. Sinha, C. Stassis, L.L. Miller and D.C. Johnston, 1990, Phys. Rev. 41, 1926. Vaknin, D., L.L. Miller and J.L. Zarestky, 1997, Phys. Rev. B 56, 8351. Vigoureux, P., M. Braden, A.G. Gukasov, W. Paulus, Ph. Bourges, A. Cousson, D. Petitgrand, J.R Lauriat,

M. Meven, S.N. Barilo, D.I. Zhigunov, P. Adelmarm and G. Heger, 1997a, Physica C 273, 239. Vigoureux, E, A.G. Gukasov, S.N. Barilo and D.I. Zhigtmov, 1997b, Physica B 234-236, 815. Wang, H.Y., C.H.Chang, S.R. Hwang, W.-H. Li, K.C. Lee, J.W. Lynn, H.M. Luo and H.C. Ku, 2000, Phys. Rev. B 62, 11549. Wu, S.Y., W.T. Hsieh, W-H. Li, K.C. Lee, J.W Lynn and H.D. Yang, 1994, J. Appl. Phys. 75, 6598. Wu, S.Y., W.-H. Li, K.C. Lee, J.W. Lynn, T.H. Meen and H.D. Yang, 1996, Phys. Rev. B 54, 10019. Xiong, Y.E, Y.S. Yao, L.E Xu, E Wu, D. Jin and Z.X. Zhao, 1998, Solid State Commun. 107, 509. Yamada, K., M. Matsuda, Y. Endoh, B. Keimer, R.J. Birgeneau, S. Onodera, J. Mizusaki, T. Matsuura and G. Shirane, 1989, Phys. Rev. B 39, 2336. Yang, K.N., J.M. Ferreira, B.W. Lee, M.B. Maple, W.H. Li, J.W Lynn and R.W. Erwin, 1989, Phys. Rev. B 40, 10963. Yildirim, T., 1999, Turk. J. Phys. 23, 47. Yildirim, T., A.B. Harris, O.E. Wohlman and A. Aharony, 1994, Phys. Rev. Lett. 72, 3710. Yildirim, T., A.B. Harris and E.E Shender, 1996, Phys. Rev. B 53, 6455. Yildirim, T., S. Skanthakumar and J.W. Lynn, 1998, private communication. Zhang, H., J.W. Lynn, W.-H. Li, T.W Clinton and D.E. Morris, 1990, Phys. Rev. B 41, 11229. Zhang, H., J.W. Lynn and D.E. Morris, 1992, Phys. Rev. B 45, 10022. Zobkalo, I.A., A.G. Gukasov, S.Yu. Kokovin, S.N. Barilo and D.I. Zhigunov, 1991, Solid State Commun. 80, 921. Zou, Z., J. Ye, K. Oka and Y. Nishihara, 1998, Phys. Rev. Lett. 80, 1074. Zou, Z., J. Ye, K. Oka and Y. Nishihara, 1999, Phys. Rev. Lett. 82, 462.

Handbook on the Physics and Chemistry of Rare Earths Vol. 31 edited by K.A. Gschneidner, Jr., L. Eyring and M.B. Maple © 2001 Elsevier Science B.K All rights' reselved

Chapter 200 HEAT CAPACITY Peter M. ALLENSPACH

Laboratory Jbr Neutron Scattering, ETH Zurich & PSI Villigen, 5232 Villigen PSI, Switzerland M. Brian MAPLE

Department of Physics University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0319, USA

Come~s List of symbols and acronyms 1. Introduction 2. Contributions to the specific heat 2.1. Electronic and lattice contribution to the specific heat 2.2. Magnetic contribution 2.3. Hyperfine contribution 3. Experimental results for Rga2Cu307 3.1. R=Y, La, Ce, Pm, Eu, Gd, Tb, Lu 3.2. Kramers ions (Nd, Sin, Dy, Er, Yb) 3.3. Non-Kramers ions (Pr, Ho, Tm) 4. Models for the magnetic ordering 4.1. Dipolar interaction 4.2. Superexchange and the RKKY interaction 4.3. Hyperfine interaction 4.4. Discussion 5. Experimental results for oxygen-deficient RBa2Cu30x

351 352 352 352 353 354 354 354 358 360 360 362

6. 7. 8.

364 366 367

9.

5.1. Introduction 5.2. PrBaeCu30 . 5.3. NdBa2Cu30 x 5.4. SmBa2 Cu30., 5.5. GdBa2Cu30 x 5.6. DyBa2Cu30~ 5.7. ErBa2 Cu3 Oy Discussion of the experimental results of RBa2Cu30~ Experimental results for RBa2Cu40 s Experimental results for R2Ba4Cu7014+x 8.1. Introduction 8.2. Magnetic specific heat 8.3. Discussion Conclusions Acknowledgment References

367

List of symbols and acronyms 2D

two-dimensional

L

total orbital momentum

3D CEF

three-dimensional crystalline electric field

R

lanthanide

J

total angular momentum

R1237 RBa2CH307 R123x RBa2Cu3 O,

kB

Boltzmann's constant

R124

35l

RBa2Cu40 ~

367 368 369 371 372 373 374 375 381 383 383 384 385 387 387 388

352

RM. ALLENSPACHand M.B. MAPLE

R247 R2Ba4Cll7014+x S spin y linear term of the specificheat

/~ /~ /~B

cubic term of the specificheat magneticmoment Bohr magneton

1. Introduction

Low-temperature specific-heat measurements are especially useful in the investigation of magnetic ordering and crystalline electric field (CEF) effects in lanthanide (R) barium cuprate high-temperature superconductors for the following reasons: Anomalies in the magnetic susceptibility due to magnetic ordering of the R ions are masked by the large diamagnetic signal in the superconducting state (i.e. below 90K). Neutron diffraction requires a great deal of effort in terms of measurement time, equipment (dilution refrigerator), and sample mass in order to investigate an entire series of samples with different oxygen contents at many different temperatures. In addition, short-range magnetic ordering as observed in R123x and R247 at certain oxygen contents yields only weak signals in neutron diffraction experiments. In this respect, an accurate knowledge of the magnetic specific heat as a fimction of stoichiometry and temperature provides an excellent basis for specific neutron investigations. On the other side, results from neutron diffraction and neutron spectroscopy (such as moment direction, size of the ordered moment, and CEF splitting) are indispensable in modeling the measured specific heat. Hence, we will often refer to ch. 194 of Vol. 30, and ch. 199 of the present volume of this Handbook series, or corresponding papers, especially to the review article by Fischer and Medarde (1998).

2. Contributions to the specific heat 2.1. Electronic and lattice contribution to the specific heat

The specific heat at low temperatures consists of electronic, lattice, magnetic, and hyperfine contributions (see for example, Junod 1996). The electronic contribution in a metal due to the conduction electrons is linear in temperature. In a BCS superconductor, the conduction electrons start to condense into Cooper pairs at Tc and the density of unpaired electrons decreases exponentially below this temperature. Hence, at the temperatures of interest in this chapter (i.e., below 4 K) the density of unpaired electrons is essentially zero as is expected for the electronic contribution. However, for RBa2Cu3Ox (R123x) a linear term of the order of y = 3-6 mJmole 1 K 2 was always observed (van der Meulen et al. 1988a,b, Junod 1996). This indicates that some conduction electrons do not pair at low temperatures, most likely due to an anisotropic or d-wave gap (i.e., with nodes). The lattice specific heat at temperatures not exceeding 10% of the Debye temperature can be approximated by fi.T 3. The Debye temperature for R123x lies between 350 and 450 K and f i = 0 . 3 - 0 . 5 m J m o l e -I K -s (van der Meulen et al. 1988a,b, Junod 1996). In most of

HEAT CAPACITY

353

the figures in this chapter the electronic and the lattice contributions (y.T+/3-T 3) have already been subtracted. 2.2. Magnetic contribution On top of the electronic and lattice contributions, and in the temperature regime up to 4 K, much stronger possible magnetic contributions either from magnetic lanthanide ordering or thermal population of CEF states are found. If neither the total orbital momentum L nor the total angular momentum J of the ground-state J-multiplet of the lanthanide ion is zero, the degeneracy of this ground state (2J + 1) is lifted by the CEF of the surrounding ions. The smallest final degeneracy for a lanthanide ion with an integer J is one (singlet), while for a half integer J (Kramers ion) the smallest degeneracy due to CEF is two (doublet). This lifting of degeneracy is called quenching of the total angular momentum and leads to a reduced moment compared to the free ion value ofgj ~/)(J + 1) gB (Jensen and Mackintosh 1991). In the absence of magnetic ordering the thermal population of the CEF-split sequence of states leads to Schottky anomalies in the specific heat (per mole of lanthanide ions) given by

where

Pi = ~1 Z

exp ( - k ~Ei) i

and

Z = ~-~ exp ( - k B ~ ) i

are the Boltzmann population and the partition function, respectively, kB is the Boltzmann constant, R is the gas constant, and Ei is the energy of an individual CEF state. For wellseparated CEF levels, the associated entropy change is

AS=R .In(g),

(2)

where t is the total number of levels involved (including degeneracy) and g is the degeneracy of the ground-state level. Hence, two doublets would yield an entropy of R In 2 which is the same as for two singlets. The entropy change can be obtained by integrating the magnetic contribution to the specific heat, Cm, over the temperature range (Ta-Tb) of the anomaly: AS =

/T;bm

~ - dT.

(3)

Equations (2) and (3) are still valid in the case of magnetic ordering, although for long-range magnetic order the broad Schottky peak is replaced by a sharp anomaly

354

RM. ALLENSPACHand M.B. MAPLE

whose form depends on the nature of the ordering. Plots of such anomalies, based on specific theoretical models (molecular field, Heisenberg, or Ising models) as well as dimensionality for spin-½ systems can be found in the review article by de Jongh and Miedema (1974). Except for the two-dimensional (2D) Ising model, the specific-heat auomaly associated with long-range ordering cannot be calculated analytically for any of the other models. Fortunately, as we will see, the character of the ordering for samples with high oxygen contents (R1237) turns out to be close to 2D Ising behavior. As can be seen in the article by de Jongh and Miedema (1974) the specific-heat anomalies of low-dimensional systems are similar to Schottky peaks. The same is true for systems which have only short-range magnetic interactions. 2.3. Hyperfine contribution The hyperfine contribution to the specific heat is observed as an upturn at very low temperatures (normally at temperatures well below 1 K) and is the high-temperature tail of a low-lying Schottky anomaly originating from a slightly split nuclear magnetic ground state. Except for oxygen, the isotopic mixture of the natural elements in the R123x compounds has an non-vanishing nuclear spin I, and hence contributes to the hyperfine specific heat. Since the splitting of the nuclear magnetic states in a magnetic field is proportional to the nuclear magneton (/~N= 3.1528 × 10-5 meV/T), its magnetic field dependence is much weaker than that for electronic magnetism (whose splitting is proportional to the Bohr magneton ktB=5.7891×10-2meV/T). Therefore, low-temperature upturns in the specific heat can easily be identified as nuclear or electronic by applying a magnetic field and observing the corresponding shifts to higher temperatures. Dunlap et al. (1987) pointed out that the nuclear magnetic hyperfine field in R1237 compounds is nearly proportional to the electronic magnetic moments of the lanthanide ions. By comparing the measured hyperfine field with the free-ion value they obtained an ordered moment for Dy and Er in R1237 which is close to the value obtained from neutron diffraction measurements (6gB instead of 7/~B for Dy and 4.5gB instead of 4.9~B for Er, respectively).

3. Experimental results for RBa2Cu307 3.1. R = Y,, La, Ce, Pm, Eu, Gd, Tb, Lu

CeBa2Cu3Ox and TbBa2Cu3Ox cannot be synthesized, while PmBa2Cu3Ox has not been made due to the short lifetime of the radioactive Pro. We could not find any information in the literature about the specific heat of LaBa2Cu3Ox. The specific heat of YBazCu3Ox is extensively discussed in the review article by Junod (1996) and references therein. The specific heats of the nonmagnetic EuBa2Cu307 and

HEAT CAPACITY

I

355

'

'

'

i

'

i

i

,

RBa2Cu307 •

0.8

R •

"7

I o

R

+. +.

0.6

+ •

O

•

0.4

+o

v• t-

v v

~l

0.21- ~

0.0

+

v

0

,,,

10

vv

. . . .

+o ° +o

.+"

''

20 r (K)

'

i

i

l

30

l

l

l

40

Fig. 1. Specific heat measured for the nonmagnetic compounds YBa2Cu307 5, EuBa2Cu307 ~ and LuBa2Cu307 6, after van der Meulen et al. (1988a). The up-turns at low temperatures are nuclear and magnetic impurity contributions.

LuBa2Cu307 compounds are qualitatively similar to that of the Y-compound (fig. 1 shows typical results for R=Y, Eu, and Lu), but the specific-heat values are higher for LuBa2Cu307. All three compounds show an upturn at low temperatures which is quite sensitive to applied magnetic fields and, hence, only a small part is due to nuclear magnetism (van der Meulen et al. 1988a,b, Kierspel et al. 1994). An obvious and very probable explanation for this upturn is the presence of magnetic impurities. There are more than twenty reports in the literature (see table 1) of the low-temperature specific heat of GdBa2Cu307 which essentially yield the same results. As can be seen in fig. 2, a sharp anomaly was observed around the ordering temperature TN = 2.2-2.3 K, and a shoulder around 1 K (especially pronounced in a C/T plot). The magnetic ground state of Gd 3+ in GdBa2Cu307 is a pure S (= 7/2) state and is therefore within first order unaffected by the CEE The degeneracy of the ground state is 2S + 1 = 8. This degeneracy is totally lifted below the magnetic ordering temperature due to the molecular field, and according to eq. (2), the entropy of the anomaly in the magnetic specific heat at the ordering temperature is expected to be R In 8. The value for the entropy obtained from the measurements is close to this value. The decrease of the ordering temperature as a function of applied magnetic field (fig. 2) is a clear indication of antiferromagnetic ordering. Hydrostatic pressure measurements performed by Bloom et at. (t987) and Alekseevskii et al. (1988) (fig. 3) yield for dTN/dp a value of 0.03Kkbar 1 and 0.007Kkbar 1, respectively.

356

RM. ALLENSPACH and M.B. MAPLE Table 1 Sources of experimental data

Publication

Rare earths

Oxygen contents"

Temperature Remarks b range (K)

Allenspach et al. (1993) Allenspach et al. (1994a) Allenspach et al. (1994b) Allenspach et al. (1995) Allenspach et al. (1996) Alekseevskii et al. (1988) Atake (1991) Basu et al. (1989) Basu et al. (1991) Bloom et al. (1987) Brown et al. (1987) Causa et al. (t987) Causa et al. (t988) Chmist et al. (1988) Dirken and de Jongh (1987) Dr6ssler et al. (1996) Dunlap et al. (1987) Dunlap et al. (1988) Escribe-Filippini et al. (1988) Ferreira et al. (1988) Gering et al. (1988) Gurevich et al. (1990) Heremans et al. (1988) Hilscher et al. (1994) Ho et al. (1987) Jana and Bhuniya (1993) Kadowaki et al. (1987) Kierspel et al. (1994)

Nd Nd, Dy Nd0 Dy Nd, Dy Sm Gd Dy, Ho, Er, Tm Ho Ho Gd Gd, Dy, Ho, Er Gd, Dy Gd Eu Dy Nd, Sm, Ho Dy, Ho, Er Gd Sm Eu, Ho, Tm, Yb Pr, Nd, Gd, Tm Tm Eu, Gd, Dy, Er Pr, Gd Gd Nd, Sm, Gd, Dy Gd Nd, Eu, Gd, Dy, Ho, Er, Tin, Yb Gd, Dy, Ho, Er Gd Gd, Ho Nd, Sm, Gd, Dy, Er Nd, Sm, Dy, Er Pr, Nd, Sin, Gd, Dy, Ho, Er Gd Gd

6 7 6-7 6 7 6-7 6-7 7 7 7 7 7 7 7 7 7 7 6-7 7 6+7 7 7 6+7 7 7 6+7 7 7 7 7

0.4~4 0.4-6 0.4-2.5 0.4M 0.4-4 2 3.5 0.2-300 70 300 1.8 12 1 5 0.3-24 0.4-7 0.5 3 2.8-100 0.5 3 0.1-6 0.120 0.1-5 2 110 0.4-30 140 2-15 22~500 1.6-70 1.8 10 100-300 1 5 0.1-8

7 7 7 7 6 7 6+7

0.1-22 0.1~4 10-300 0.4 30 0.4-30 0.4 24

6+7 7

1.5 3 1.5-5

Kobayashi et al. (1987) Kohara et al. (1987) Lazarev et al. (1988) Lee et al. (1988) Lee et al. (1990) Maple et al. (1989) Meyer et al. (1987) Mori et al. (1988)

Ca doping, Ndl+yBa2 ~Cu30~ Magnetic field Magnetic field, pressure Pressure

Magnetic field

(Eul .yGdy)Ba 2Cu 307

Magnetic field

Sr doping Magnetic field, Sr doping

Magnetic field Magnetic field

Magnetic field (Yi ~Prv)Ba2Cu307

c o n t i n u e d on n e x t p a g e

HEAT CAPACITY

357

Table 1, continued Publication

Rare earths

Oxygen contents"

Temperature Remarks u range (K)

Nakazawa et al. (1987)

Gd, Dy, Ho, Er, Tm

6+7

0.4~6

Naumov et al. (1996) Niraimathi et al. (1994)

Ho Nd

7 7

10-300 0.5-5

Niraimathi et al. (1995a)

Nd

6+7

0.5-4

Niraimathi et al. (1995b) Oota and Kojima (1988)

Nd Gd

6.85-7 6~

0.5-60 1.8-30

Oota et al. (1989)

Gd

6-7

1.5~2

Fe doping

Plackowski and Wlosewicz (1996) Ramirez et al. (1987)

Er

7

10-300

(Yi yEry)BaeCn; O7

Pr, Nd, Sin, Gd, Dy

7

0.4~10

Magnetic field

Ravindran et al. (1995)

Nd

7

~50

Zn doping

Reeves et al. (1987)

Gd

7

1.5-180

Schaudy et al. (1992) Schaudy et al. (1993)

Pr, Gd Pr

7 6+7

1.8-25 1.8-24

Simizu et al. (1987a)

Nd, Sm, Gd, Ho, Er

7

0.4-20

Simizu et al. (1987b)

Gd, Ho, Er

7

0.4 100

Ga doping Ga doping Ga doping Fe doping

Magnetic field Magnetic field, (YI yP~)B~Cu307

Simizu et al. (1989)

Er

6-7

0.1-7

Slaski et al. (1988) Thuy et al. (t992)

Gd, Dy, Ho, Er Er

6+7 7

0.l-20 2-40

Uma et al. (1996) van den Berg et al. (1987)

Pr Gd

7 7

1.8~5 2-32

Magnetic field, single crystal

van der Meulen et al. (1988a) van der Meulen et al. (1988b)

Pr, Sin, Eu, Gd, Dy, Ho, Er, Tm, Yb, Lu Pr, Dy, Yb

7

1~40

Magnetic field

7

1~5

Magnetic field

Wang et al. (1989) Willekers et al. (1990a)

Gd Tm

7 7

80-90 0.05-0.8

Willekers et al. (1990b)

Yb

7

0.1-1

Willekers et al. (1991)

Ho, Er, Yb

7

0.05 1

Willis et al. (1987) Yang et al. (1989)

Gd Nd, Sm

7 6+7

1.5~5 0.4~4

Zhu et al. (1988)

Gd

6+7

1.8-4.5

7 means that a sample with an oxygen content close to 7 was investigated; 6+7, one close to 6 and one close to 7 was investigated; 6-7, a series of samples with different oxygen contents. b Additional measurements and parameters.

358

RM. ALLENSPACH and M.B. MAPLE

16.0

.... i .... i ~_, "-'t-JctBa2t--u3u7 "" "-" .

14.0

~d '~

+

12.0

.... i , , , Brown Bloom: 0Testa Bloom: 1.7 Tesla Bloom: 3.5 Tesla

++

10.0

~-

~-+

8.0 [.., "~ Q) <~

, i + • • •

~

6.0

+ +++++-~ +

4.0

+

+

"++ "_+

*

; • --'+ ~,+

o

..

0.0

.... 0

J .... 1

I . 2

.

; ;; .

. 3

.

,

4

5

T (K) Fig. 2. Magnetic specific heat data for GdBa2Cu307 ,~ with and without applied magnetic field according to Brown et al. (1987) and Bloom et al. (1987).

3.2. Kramers ions (Nd, Sm, Dy, Er, Yb) RBa2Cu307 compounds with a Kramers ion at the R site have been studied extensively in the past (table 1). The results are essentially all in agreement, with the following two exceptions: First, early specific-heat measurements on NdBa2Cu307 do not show a sharp transition at the ordering temperature due to sample inhomogeneity (Ramirez et al. 1987, Simizu et al. 1987a). Second, magnetic features with entropies of the order of 1 Jmole 1 K-1 have been observed at 5.5 K for SmBa2Cu307 (Escribe-Filippini et al. 1988), at 11K for DyBa2Cu307 (Lee et al. 1988, van der Meulen et al. 1988a,b), at 5 K for ErBa2Cu307 (Brown et al. 1987, Dunlap et al. 1987, Simizu et al. 1987a,b, Slaski et al. 1988, van der Meulen et al. 1988a,b), and at 2 K for YbBa2Cu307 (Ferreira et al. 1988, van der Meulen et al. 1988a,b), respectively. All of these features disappeared with better sample quality in subsequent studies. Figure 4 displays typical specific-heat anomalies of RBa2eu307 (R=Nd, Sm, Dy, Er, Yb) due to magnetic ordering. The ordering temperatures are 0.52, 0.61, 0.91, 0.60, and 0.26 K for Nd, Sm, Dy, Er, and Yb, respectively. Specific-heat measurements in an applied magnetic field (see, e.g., Kierspel et al. 1994) as well as neutron diffraction experiments (see ch. 199 of this volume) proved the ordering to be antiferromagnetic. For Dy, Er, and Yb, the anomalies are very sharp and almost symmetric, while for Nd and to a lesser extent for Sm there is a sharp cusp at TN and a broad high-temperature tail. From neutron spectroscopy the

HEAT CAPACITY

359

35 30

)

25

~

15

/JI

.....,kbar,

10 5 I

02

I

I

I

]

I

I

~

I

I

2.5

I

n

I

I

3

3.5

T (K) Fig. 3. Pressure dependence of the magnetic ordering of GdBa2Cu307 (Alekseevskii et al. I988). The lines are guides to the eye.

15 RBa2Cu3OT_ ~ ,

•

Vv

.~ 10

Nd

~+

•

•

O •

5 <1

A/t

+

A

+

,

,,

~.+

•

*~+

+

"%+*++

+

;K..++

~t

t

,~,

I~'

0

o

0

I

,

a

a

-I

0.5

t

n

, ,

•

~,, n

T

n

•

"'...: 2 + + j + ,

,_

"" "":. t ....+. ~.-

P

1

n

u

l

n

]

1.5

n

a

n

n

2

T(K) Fig. 4. Magnetic specific heat of NdBa2Cu307 6 (Maple et al. 1989), SmBa2Cu307 6, DyBa2Cu~O7 ~ and ErBa2Cu30 ~ ~ (Lee et al. 1988), and YbBa2Cu307 ~ (Willekers et al. 1990b).

360

RM. ALLENSPACHand M.B. MAPLE

CEF level schemes for these four compounds are well known: The first excited states lie above 10meV for Nd, Sm, Er, and Yb and hence do not contribute to the specific heat below 4 K (see Vol. 30, ch. 194 of this Handbook). For Er, the specific heat calculated from CEF parameters obtained by neutron spectroscopy were compared to measured data at 0 and 5 tesla by Mesot et al. (1993a). The first excited level for Dy lies at 3.3 meV and, as a result, the low-temperature tail of the corresponding Schottky anomaly is observed down to about 2 K (see, e.g., Dunlap et al. 1987) and has to be taken into account in the analysis of the ordering peak. Pressure measurements performed on NdBa2Cu30697 and DyBa2Cu306.95 yield dTN/dp=0.0049 and 0.0035 Kkbar -j, respectively (Allenspach et al. 1995). Since the ordering temperatures for these two compounds (and GdBa2Cu307, see previous paragraph) are not the same, a comparison of the values is only reasonable by taking the relative changes (dTN/TN)/dp. The results are: 0.0094 (Nd), 0.0054 (Dy), and 0.0031-0.0133 Kkbar 2 (Gd). 3.3. Non-Kramers ions (Pr, Ho, Tin) In contrast to Kramers ions, non-Kramers ions do not automatically have a magnetic ground state. Here the low-temperature magnetic behavior is directly controlled by the CEF splitting. For PrBa2Cu307 and HoBa2Cu3OT, the ground state is formed by a quasi triplet and a quasi doublet, respectively. Induced magnetism results in magnetic ordering at 17 K for Pr and 0.17 K in the case of Ho. TmBa2Cu307 has a well-isolated singlet ground state (see Vol. 30, ch. 194 of this Handbook) and does not order at all. The magnetic specific heats of the three compounds are displayed in fig. 5. Magnetic contributions at elevated temperatures have their origin in the population of higher CEF states. The Schottky specific heats calculated from the observed CEF level splitting measured with neutron spectroscopy are given by lines in fig. 5. The disagreement is mainly due to an inaccurate subtraction of the electronic and lattice part. PrBa2Cu307 is an exception for many reasons: It is not superconducting l, has an ordering temperature one order of magnitude higher than the rest of the lanthanides, and has no sharp CEF transitions. All this indicates strong hybridization with the surrounding oxygen and Cu orbitals.

4. Models for the magnetic ordering The specific-heat observation of magnetic ordering (except for R = E u , Tm, Lu) in RBa2Cu3 07 was confirmed by neutron diffraction, and the ordering temperatures obtained by the two methods are in agreement. Very early in the specific-heat investigations in these compounds, the similarity of the measured anomalies at TN with the analytical result of i Superconductivitywith Tc ~ 85 K has been reported in P r g a 2 C u 3 0 x samplespreparedby the traveling-solvent floating zone method (Zou et al. 1998). These materials show large inhomogeneitiesin both lattice parameters and transport and magnetic properties. In parts of the sample that exhibit superconductivity,the c-axis lattice constant is reported to be larger than that of the nonsuperconducting ceramicpolycrystallineor single-crystal samples prepared by standard methods.

HEAT CAPACITY

20

361

,,,l,,,,l,,~,l,,,,l,,,,l,,,,l,,,,l,,,,

RBa2Cu307_ ~ 15

15 "7q~ O

1

ii.

Ho

•

,

"

% ~"

•

,

¢

vt

¢

•

Pr

•

Ho

Ho-CEF

+,m

.........

"',.-.~

++++

0.4

+++

10

/ ~00000•

<1

.0

5

+++

÷÷++

++

/_.,:.+'+'++++++++++ +Jr+++

++++++++++++

-~t~l

00

Tm-CEF

~-, , , , I ....

5

10

........

.L.H--C:'-[',

15

pd

"

8D f ~ B f ~

. - -"""

......

, , , I , , ,

20 25 T (K)

I , , , , I , ,

30

35

,

40

Fig. 5. Magnetic specific heat of PrBa2Cu307 ,7 (Urea et al. 1996), HoBa2Cu307 ~ (Maple et al. 1989; inset: Willekers et al. 1991), and TmBa2Cn30v ~ (Gering et al. 1988). The CEF contributions were calculated from neutron spectroscopy data (see Vol. 30, ch. 194 of this Handbook). Table 2 Magnetic exchange constants obtained by a pure 2D Ising fit Rare earth Nd Sm Gd Dy Er Yb

Jl (~eV)

J2 (~teV)

59 -60 168 -63 47 - 17.4

3.5 -5.4 45.7 -15.5 8.6 4.3

Reference Allenspach et al. (1994a) Lee et al. (1990) Simizu et al. (1987a) Lee et al. (1990) Lee et al. (1990) Willekers et al. (1990b)

the 2D Ising m o d e l (Onsager 1944) was recognized by Simizu et at. (1987a,b). The inplane coupling constants obtained by such a 2D Ising calculation are listed in table 2. Lee et al. (1990) pointed out that the anisotropy in the coupling constants in the p l a n e necessary to explain the experimental data is lanthanide dependent: J1/J2 = 50, 1 t, 4, and 5 for Nd, Sm, Dy, and Er, respectively. These large anisotropies are surprising, in v i e w o f the fact that the orthorhombicity in these compounds, even at full oxidation, is n o t very large and that the orthorhombicity for the light lantbanide ions is the smallest. The

362

RM. ALLENSPACH and M.B. MAPLE

2D Ising model is based on nearest-neighbor interactions only, favoring superexchange and, to a lesser extent, the RKKY interaction over the long-range dipolar interaction. The usual way to distinguish exchange and dipolar interactions is to plot TN as a function of the de Gennes factor gj 1)2.J(J + 1) or as a function of the square of the ordered moment, respectively. These two types of plots are displayed in fig. 6. One problem with the de Gennes scaling is that at low temperatures where the ordering occurs, the whole ground-state J-multiplet is not thermally populated due to the CEF splitting. This leads to a quenching of the total angular momentum and hence the J in the de Gennes factor loses its meaning. We can artificially compensate for this effect by using ~u=gj V / ~ J + 1) and inserting the ordered moment ~ord for/~. The alternative de Gennes factor is equal to (1 -- [email protected])2'~2rd (see fig. 6c). The ordering temperatures follow quite nicely the de Gennes scaling but, nevertheless, we cannot rule out the possibility that the dipolar interaction plays a role, since it is always present.

4.1. Dipolar interaction We start with a calculation of the ordering temperature solely based on the dipolar interaction. According to calculations by Smit et al. (1987) the contributions to the dipolar fields due to adjacent R layers are insignificant; hence, a 2D calculation is sufficient to obtain an idea of the importance of the dipolar interaction. For our cluster calculations (following the method described by MacIsaac et al. 1992), we used the ordered magnetic moment obtained from neutron diffraction or - if no data were available - the value calculated from the CEF parameters. The size and direction of the moment together with the lattice constants used for our calculations are listed in table 3. The Hamiltonian for dipolar interaction is given by

• ~."

r 5

'

(4)

with/7/the magnetic moment at site i and F/j the direction vector from site i to site j. For the magnetic ordering in R B a 2 C u 3 0 7 we will, for simplicity, assume that the moments are parallel or antiparallel (Ising spins) to each other and the size of the moment is not site dependent. (We will see in a later section that for oxygen-deficient PrBa2Cu3Ox and NdBa2Cu30~, the moments are tilted.) The Hamiltonian (4) then simplifies to

(5) i ~'j

~

where 0,..1.is the angle between F/j and the (fixed) moment direction. With the values listed in table 3 the unit of dipolar energy i s / ~ / ~ 3 = 5.368× 10 .2 meV = 0.623 K. The ordering temperature was obtained by calculating the specific heat and determining the position of the maximum of the anomaly. The specific heat was the result of a Monte-Carlo evaluation

HEAT CAPACITY 2.5

'

'1

'

'

'F

'

'

' t ' '

2.0 (a)

'l

'

'

"

'1

'

,

,I

,

,

,i

,

,

363

,

~

1.5 z

1.0

evY

0.5

No °~

0.0;

"2'

s.,

4' " ' ' ' '6' " " ; "8' [ 2 1, , , 1 J14 (gj- 1)2-J(J+l)

2.5 . . . . . . . . (b)

i ....

I ....

,, 16 I .... %.

2.0 1.5 Z

1.0

by+

0.5 oN.

. . . .

+o

(gord)2

2.5

' ' 1

'

' ' l ' ' ' l '

' ' l ' '

' 1 '

(c)

''

~

2.0 1.5 ~z 1.0

iDy

0.5 f~

0"~0

I

2

4 6 8 (1- l/gj)2. (~.[ord)2

10

12

Fig. 6. Magnetic ordering temperatures as function of (a) the de Gennes factor, (b) the moment squared, and (c) the modified de Gennes factor.

364

P.M. ALLENSPACHand M.B. MAPLE

Table 3 Measured in-plane lattice constants, size and direction of the magnetic moments, ordering temperatures used for the calculations~, and calculated values for the ordering temperature (with dipolar interactionsonly) and the magnetic exchange constants(dipolar + exchange interactioncalculations) Rare earth

a (A)

b (A)

]#l (/~B) #-direction Te~ N (K)

T~"I~ (K)

J (~teV) Jcalc

Jcorr b

Pr123 Nd123 Sm123 Gd123 Dy123 Ho123 Er123 Yb123

3.86 3.86 3.84 3.83 3.82 3.82 3.81 3.80

3.92 3.91 3.90 3.89 3.89 3.88 3.88 3.87

0.56 1.14 0.34 6.9 6.8 2.8 4.9 1.44

tilted c a c c a b b

17 0.52 0.61 2.25 0.91 0.17 0.60 0.26

<0.02 <0.02 <0.02 0.62 0.64 0.23 0.61 0.06

650 35000 -59/-3.5 -323/-19.2 60~5.4 83./ 7.5 -58/--58 6.2/ 6.2 8~8 -2.8/-2.8 -12/-12 -18.0/-18.0 17.4~4.3 537/-133

Dy124

3.84

3.87

5.9

c

1.00

0.47

-21/-21

-7.4/-7.4

Dy247 Er247

3.83 3.83

3.87 3.86

7 3.68

c b

1.00 0.54

0.64 0.35

-12/-12 -27/-27

-4.2/-4.2 41/ 41

'~ Values hardly differ from author to author. b Exchange constantscorrected with the modifiedde Gennes factor (see text). of C = /q~((Edip) 1 2 - (Edip) 2) (Yeomans 1992), where we started at low temperature with the moment arrangement found by neutron diffraction. The rare-earth cluster used for the calculation had a size of 10× 10 ions. The eight adjacent clusters where included in the calculation in order to simulate the long-range character of the dipolar interaction. The spin arrangements of these patches were identical to the central one. Since this calculation should just provide an approximation, only 105 Monte-Carlo steps were calculated. The results of these calculations are listed in table 3. Obviously, dipolar interaction alone cannot account for the magnetic ordering temperatures, except for Ho and Er. For Dy the dipolar interaction is dominant, for Gd it is significant, and for Pr, Nd, Sm and Yb it is almost negligible. The form o f a calculated specific-heat anomaly at the ordering temperature based solely on dipolar interaction (fig. 7) turns out to be similar to a 2D Ising curve (especially the high-temperature tail), hence the large anisotropy obtained for the coupling constants for the 2D Ising model might, for Dy or Er, have its origin in the presence of the dipolar interaction. However, this does not solve the problem of the even larger anisotropy for Nd or Sm where dipolar interactions are negligible. 4.2. Superexchange and the RKKY interaction The good de Gennes scaling of the ordering temperatures was always regarded as a proof of the importance o f the RKKY interaction despite the fact that all other experimental results did not support such an interpretation: With lSSGd M6ssbauer spectroscopy

HEAT CAPACITY 10

,,,,,,,,,,,,,,

365

I l l l [ ~ l l * [ l ~ l l [ l l l l ] l l l *

- - d i p o l e interaction g = 7 gB (//C) a = 3.82 A; b = 3.89 ,~ ..... exchange interaction J~ = -42 geV; J2 = -10.3 ~teV

© 0

o

0

0.5

1

1.5

2 2.5 3 3.5 4 r (K) Fig. 7. Comparison of a specific heat anomaly calculated for Dy1237 with pure dipolar interaction and one calculated with a anisotropic exchange interaction (for both calculations a square of 10× 10 magnetic ions was used)• Since the dipolar interaction and the 2D Ising model belong to the same universality class it is not surprising to find exchange values which exactly reproduce the dipolar data. (Smit et al. 1987), ESR (Causa et al. 1987) and N M R (Alloul et al. 1988), the density of charge carriers at the R site was found to be extremely small. By means of specificheat measurements, the ordering temperatures in oxygen-reduced (insulating) samples (see sect. 5) were shown to be almost identical to those in the fully oxidized (metallic) samples (except for Pr, Nd). The transformation of an exchange Hamiltonian into the Heisenberg spin Hamiltonian (which results in the dependence of the ordering temperature on the de Ge~mes factor) is independent of the type o f exchange interaction, be it R K K Y or superexchange. For the reasons mentioned above, we believe that the superexchange interaction is - at least in the low-oxygen regime - more important than the R K K Y interaction in these compounds. The superexchange interaction is mediated by the oxygen in the CuO2 planes. Since the holes responsible for superconductivity are planar oxygen holes, some influence of the electronic properties within these planes on the magnetic ordering of the lanthanides is expected. In order to include the spin-½ exchange interaction into our dipolar model, we extended the Hamiltonian (5) to

~r: Z •

.~ j

-t-ZS-(1 .

r(/

l

3cos0/j)-2ZJa(S[.ST),

(6)

(i,.i)

where J a is the exchange coupling in one o f the two plane directions and (i,j) is the sum over neighboring pairs. The summation for the exchange part was only performed

366

RM. ALLENSPACH and M,B. MAPLE

40

''''

I''''

I ''''1

''''

dipolar and exchange interaction

35

..... Gd1237 I ~Dy1237 I ---Er1237 I

30 "T"

O

25 20

,,

10

~ I

5

,'"

I • ,1

0

0.5

• o.

...." ."

"''""

.....

,"

1

1.5

2 2.5 T (K)

3

3.5

4

Fig. 8. Calculated specific heat anomaly (dipolar and exchange interaction; cluster size: 10×10) for GdBa2Cu~O7, ~, DyBa2Cu307 _ ,~ and ErBa2Cu3OT_ d. The data for Dy and Er are very similar to the experimental data (see fig. 4), while for Gd the shoulder of unkalown origin at l K is not reproduced (fig. 2).

in the central 10× 10 spin cluster including the boarders (periodic boundary condition). The results of these calculations are shown in fig. 8 for the lanthanides with a strong dipolar contribution (the specific heat for the others can be modeled with the 2D Ising model alone). The coupling constants used for the calculation were chosen in order to reproduce the measured ordering temperature. In the exchange part ofeq. (6), the modified de Gennes factor should actually appear, since we transformed the problem into spin space. We did not include this factor in order to obtain the raw coupling constants which can directly be related to the ordering temperatures during the calculation, but we listed the raw and corrected coupling constants in table 3. All the calculated anomalies are due to the limited size of the cluster, broader than the measured anomalies, but qualitatively they agree. For two compounds, namely Pr and Gd, the calculation based on a spin-½ model is obviously incorrect, since the ordering of Pr involves three levels (effective spin-1) and Gd is a spin -7 system. But it was mentioned by van den Berg et al. (1987) for Gd that the shape of the anomaly is very similar to that of the spin-½ model. But according to eq. (2), the size of the anomaly has to be scaled (R in 8, instead of R In 2, for the entropy; hence the calculated anomaly has to be multiplied by a factor of 3). 4.3. Hyperfine interaction It was shown by Roessli et al. (1993a) that the induced magnetic ordering of Ho is driven by the hyperfine coupling which turned out to be 39 mK. The exchange constants obtained by Guillaume et al. (1994b) are Ja =-Jt, = 1.3(2) geV. A sizeable influence of

HEAT CAPACITY

367

the hyperfine interaction on the magnetic ordering could not be found for any of the other compounds. 4.4. Discussion In the preceding sections (4.1-4.3), we were able to show how the dipolar and exchange interaction together are responsible for the magnetic ordering in RBazCu3OT. However, to achieve this result we had to make some simplifications: pure 2D ordering, Ising spins (spin-½), and very limited cluster sizes. Just from structural aspects and, in addition, from measurements of spin excitations in grain-aligned HoBa2Cu307 (Staub et al. 1993), we know that the magnetic coupling along the c-direction is very weak. Nevertheless, in all Rga2Cu307 compounds, three-dimensional (3D) ordering was found by neutron diffraction. Ising spins may be justified in the case where the single-ion anisotropy due to the CEF is large (Dy), but for lanthanides which are almost isotropic (Nd, Gd, Yb) or rare earths which possess an easy plane (Sm, Er) the question of why the agreement of the calculations with the measurements is nevertheless quite good remains unanswered. The limitation to small cluster sizes results in some quantitative, but not really qualitative, differences, i.e. the anomalies are broader and not as high as for larger clusters and the infinite 2D Ising model. An inclusion of 3D ordering and an increase of cluster size essentially depends only on the power of the computer used for the calculation, while a more general choice of the spins (e.g., Heisenberg spins) requires a much more elaborate quantum-mechanical Monte-Carlo routine (which is normally restricted to rather small clusters). The choice of antiferromagnetic exchange constants is the result of observations of the magnetic ordering in systems with small influences from the dipolar interaction and the consideration of consistency (same sign for the exchange interaction for all of the lanthanides): For Nd and Gd where the dipolar interaction is weak (hence the magnetic ordering is controlled by the exchange interaction), antiferromagnetic ordering within the plane has been observed by neutron diffraction (Fischer et al. 1989, Chattopadhyay et al. 1988). The corrected exchange constants in table 3 should be equal for all of the lanthanides if their exchange coupling is the same. Obviously, the larger lanthanides are coupled more strongly than the smaller ones, which is due to a increased hybridization. The only exception is Yb, the smallest lanthanide ion in our investigation. Due to the limited amount of experimental data for Yb, a possible influence of the hyperfine interaction on the magnetic ordering cannot be ruled out.

5. Experimental results for oxygen-deficient RBazfu3Ox 5.1. Introduction

Oxygen reduction in R123x leads to a decrease in Tc which is dependent on the lanthanide (Buchgeister et al. 1990): For light lanthanide ions the decrease is rather rapid, while

368

RM. ALLENSPACH and M.B. MAPLE

for the heavier ions the superconducting regime extends down to low oxygen contents. This larger range of superconductivity is mainly due to the occurrence of the plateau structure of Te(x) observed for the heavier lanthanides. At a certain oxygen content (about 6.6 for Nd and 6.35 for Er), a transition from metallic to semiconducting behavior is observed. The reduction of the oxygen content causes a reduction of the mobile charge carriers (holes) in the CuO2 planes, which is well documented by transport measurements. Such a variation of the carrier density is not expected to affect the dipolar interaction directly, but it may be influenced via an oxygen dependence of the lattice constants or the CEE The former effect modifies the dipolar energy proportional to r 3, while the latter may alter the size of the moment. (With neutron diffraction and CEF spectroscopy, it was shown that the moments in oxygen-reduced samples are slightly smaller than in fully oxidized samples.) These modifications in the dipolar interaction can be calculated directly and hence will not yield unexpected results. The RKKY interaction, which is an indirect exchange via the conduction electrons, will vitally depend on the density of charge carriers. Since the fully oxidized samples exhibit metallic behavior and the reduced samples display semiconducting behavior, the RKKY interaction is expected to weaken drastically as the oxygen content decreases from x = 7 to x = 6. A semiconducting or insulating matrix is ideal for the superexchange interaction; therefore, superexchange (if present) is expected to be strong in the reduced samples. Slight doping of such a magnetically ordered semiconductor might lead to a behavior described in a paper by Zener (1951) and in a comprehensive review article by de Gennes (1960). The main result of this so-called "double exchange" would be a simultaneous observation of ferromagnetic and antiferromagnetic neutron peaks and a canting of the magnetic moments. This theory was developed for 3d magnetic ions, and it is unclear if it could play a role in 4f systems. The expectation for higher carrier concentration is a superexchange influenced by a sea of free carriers and hence a possible competition between superexchange and RKKY interaction as described by Gon~alves da Silva and Falicov (1972). Carrier-density variations in the sample will then obviously have a strong influence on the magnetic ordering if both interactions become similar in strength.

5.2. PrBa2Cu.~Ox Measurements for Pr123 with x = 7 and x = 6 were made by Gering et al. (1988), Schaudy et al. (1993), and Hilscher et al. (1994). They all observed a decrease of the magnetic ordering temperature from 17 K to 11 K in going from the fully oxidized to the reduced sample (fig. 9). No samples with intermediate oxygen contents have been measured up to now.

In the whole range of oxygen content, Pr samples are insulators with a strong hybridization of the 4f bands with the 2p oxygen and the 3d copper bands. Boothroyd et al. (1997) have shown - b a s e d on neutron single-crystal diffraction - that the magnetic moments (/~=0.56/~B for x - 7 and 1.15~B for x = 6 ) are canted.

HEAT

16

'

14

'

'

8

~

6

I

'

'

'

369

'

I

. . . .

I

. . . .

PrBa2Cu3Ox

•

~'~ ~"~

'

CAPACITY

X

."

UI

.',"*T,

, i ~

.,.v

,.,............"'""

~

4 2 0

'

0

,

5

10

,

,

,

I

15

,

,

,

,

I

,

,

,

,

20

25 W(K) Fig. 9. Specific heat of PrBa2Cu3Q for two samples with oxygen contents close to 7 and 6, respectively (Schaudy et al. 1993). 5.3. NdBa2Cu~Ox The low-temperature specific heat of reduced Nd123 was reported by Maple et al. (1989) and Yang et al. (1989). They observed a strong anomaly with a rounded maximum at 1.5 K in a non-superconducting sample. Investigations of seven samples with oxygen contents between 6.93 and 6.26 by the same group (Lee et al. 1990) yielded the observation of the most dramatic dependence of the magnetic ordering on the oxygen content (fig. 10): For very high oxygen contents (where long-range magnetic ordering still persists), the anisotropy Jl/J2 obtained from the 2D Ising fits increases from 14 for x = 6.97 to 100 for x = 6.90 (Allenspach et al. 1996; fig. 11). Just below an oxygen content of 6.9, the longrange magnetic order collapses and only a broad peak in the magnetic specific heat can be observed. By further lowering the oxygen content to 6.6 this peak shifts to higher temperatures (for 6.64, the peak maximum is at 1.4 K). The shape, which is Schottky-like, remains almost unchanged, except for a small increase of the maximum value between 6.8 and 6.7. Then by entering the non-superconducting regime below 6.6, this feature starts to sharpen again and at 6.26 becomes an anomaly indicative for long-range 3D magnetic ordering at 1.65 K. This ordering could be shown to be antiferromagnetic by measurements of the magnetic-field dependence of the specific heat (Allenspach et al. 1995) and by neutron diffraction (Clinton et al. 1993). From the neutron diffraction measurements it could be concluded that the magnetic moment is canted by 45 ° relative to the c-direction and the moment is reduced compared to the moment in the fully oxidized sample. Measurements of the magnetic specific heat in Ndj +yBa2_ yCu3Ox (Allenspach et al. 1993; fig. 12) where the additional Nd occupies the Ba sites and acts as an electron donor

370

P.M. ALLENSPACH and M.B. MAPLE

NdBa2Cu30 x

12

,

o x=6.97 o x=6.90 x=6.83 o x=6.80 x=6.72 x=6.64 • x=6.52 • x=6.43 ° x=6.26

:

-"

2.~ °

10

#

o

•

8

Oo

°

oo

6

•

°

/

."

l°

,"

,

,2.: •

x~6.02

4

0.5

1

1.5 T(K)

2

2.5

Fig. 10. Oxygen dependence of the magnetic specific heat of NdBa2Cu3Ox (Allenspach et al. 1995). Superconducting samples are indicated by open symbols.

.--.. 10 "7

i .....

8

[~ I~

M

x = 6.9:/ . . . . J = -0.068 meV J'=-0.005 meV

t

....

~ ~ .4

x= 6.96 .... = J -0.073 meV J = -0.004 meV

2 0

/ . . . . . . . . . . . . . . . . . . . . . . . . . . .

10 ~"

"7

4

"~ <1

2

J =-0.101 meV J =-0.001 meV

J = -0.084 meV J = -0.002 meV

6 o

x = 6.90

x = 6.93

8

0 0

.L 0.5

1 T (K)

1.5

2

0

0.5

1

1.5

2

T (K)

Fig. 11. Oxygen dependence of the magnetic specific heat ofNdBa2Cu30 ~ for high oxygen contents (Allenspach et al. 1995). The lines are results of 2D Ising calculations with the coupling constants given in the individual figures.

HEAT CAPACITY

14

Ndl+yBa2_yCU30 x

12

•

• "-"

~ • • ,~"I~* °

y=O.25/x=6.98 y=O.25/x=6.50 y=O.50/x=6.98 y=0.50/x=6.55

J'-'

10

%

371

8

d

6

0 ~ 0.5

1

/i

1.5 T [K]

A

2

2.5

Fig. 12. Oxygen and Nd-doping dependence of the magnetic specific heat of Nd 1+eBa2 y Cu~ 0 al. 1993). The superconducting sample is indicated by open symbols.

x

(Allenspach et

(reducing the number of holes and destroying superconductivity) have proven the magnetic behavior of Nd123 to be solely controlled by the charge-carrier density within the CuO2 planes. Low-temperature specific-heat measurements of NdBaCu3 yGayO~ (where Ga similar to the oxygen reduction process reduces the hole concentration and depresses To) show a dependence of the magnetic ordering as a function of Ga content which is very similar to that of the undoped compound as a function of oxygen content (Niraimathi et al. 1994, 1995a,b). 5.4. SmBa2Cu30x Magnetic specific-heat data for reduced SmBa2Cu3Ox have been reported early by our group (Maple et al. 1989, Yang et al. 1989). Instead of the sharp anomaly observed for the fully oxidized compound, a broad feature peaking at around 1 K was obtained for the reduced sample. A more detailed investigation (Allenspach et al. 1996) revealed a behavior of the magnetic specific heat as a function of oxygen content similar to NdBa2Cu3Ox (fig. 13): By going from x = 7 to x = 6 , the in-plane anisotropy increases and long-range magnetic order is lost below x = 6.9. Between x = 6.9 and x = 6.4 only broad (Schottky-like) features, as in the earlier papers, were observed. Below x = 6.4, the magnetic signal sharpens again with the maximum at 0.7 K. To our knowledge, no neutron diffraction measurements of the magnetic ordering of Sm in SmBa2Cu3Ox have been performed. Hence, the assumption of a re-appearance of long-range magnetic order

372

P.M. ALLENSPACH and M.B. MAPLE 2

E

. . . . o

i

i

. . . .

r

SmBa2Cu3Ox

o

10

. . . .

~, x

6.75 I

~ x

6.50 I

x

6.27 I

- x

6.13 I

•

oo

"7

.

o x = 6.98 I x 6.90 I

o

8

,

o

o

-""

,

o

0

o

~•

6 4

8

o° ~ . o °. off

~o~oX[]~

2

~ o.~,~o

° []

oo" ouo;,

0

025

. . . .

i

. . . .

1.'5'

~

:

'~

u

O O o o o ~m o[]~ "i

. . . .

2.5

T (K) Fig.

13,

Oxygen

dependence

o f the m a g n e t i c

specific

heat o f S m B a = C u 3 0 ~ ( A l l e n s p a c h

et al.

1996).

S u p e r c o n d u c t i n g s a m p l e s are i n d i c a t e d b y o p e n s y m b o l s .

at low oxygen contents is only based on these specific-heat data and the similarity with other systems such as Nd, Dy, and Er. The disagreement of our data with those published by DrSssler et al. (1996) seems to be due to the poor oxygen control of their sample preparation technique. (The same effect can be observed for Nd.) 5.5. GdBa2Cu30x The absence of almost any oxygen dependence of the magnetic ordering of Gd in GdBa2Cu3Ox observed with specific-heat measurements (Meyer et al. 1987, Nakazawa et al. 1987, Dunlap et al. 1988, Gering et al. 1988, Slaski et al. 1988, Zhu et al. 1988, Hilscher et al. 1994) was taken as one of the key arguments for a clear separation of the magnetic and the electronic sublattice and for the negligible influence of the RKKY interaction (since one sample is metallic and the other semiconducting). Nevertheless, neutron diffraction measurements revealed a dependence of the magnetic ordering on the oxygen content of the sample: Chattopadhyay et al. (1988) observed antiferromagnetic ordering along the c-axis in fully oxygenated GdBa2Cu3Ox, while this coupling was found to be ferromagnetic for a non-superconducting sample. Guillaume et al. (1994a) interpreted their data for a sample with x = 6.1 as a mixture of antiferromagnetic and ferromagnetic coupling (simultaneous appearance of the propagation vectors rL L 11 and L2~2~2J 1 1 [ 3, :,0])" The specific-heat data of Oota and Kojima (1988) and Oota et al. (1989) suggest an oxygen dependence of the specific-heat anomaly (fig. 14). Unfortunately, the number

HEAT CAPACITY

35

'

J

,

,

[

, ' ' '

I ' ' ' '

I ' ' '

'

I

GdBa2Cu30×

'

'

373

'

'

I

'

'

'

'

I

'

'

'

'

I

'

'

'

'

ooo o

30

o ortho-I [ ortho-II [

.-.

o o o

• tetra-II [

. q, ;

o

25 0 •

0

20

.

0

o 2

0 A

I 1 ~o

"~

• o ~ • 0 0 0 ~ ~x • o • io00<~

15 •

O°

n~ ~A ~x

A

~

~oO% lO

, , , F , , , , I , , , , I , , , , I ....

1.8

1.9

2

2.1

2.2

i ....

2.3

i , ,

2.4

o,^,

2.5

2.6

T (K) Fig. 14. Oxygen dependence of the magnetic specific heat of GdBa2Cu30, (Oota and Kojima 1988). Superconducting samples are indicated by open symbols.

of measured samples (3) is too small to see any decay of the long-range order which is expected between the O-II and the T-II sample. Here further investigations on samples with different and well-controlled oxygen contents are needed. 5.6• DyBa2Cu~O~ The first measurements of an oxygen-deficient sample of Dy123x by Nakazawa et al. (1987) resulted in an anomaly at about 0.8 K which is less intense and broader than that of the fully oxygenated sample (with TN=0.9K). The authors could successfully fit these data with a spin-½ XY model and suggested that for the x = 7 sample, a uniaxial single-ion anisotropy in the plane (responsible for the Ising behavior) becomes more isotropic in the non-superconducting samples. Measurements by Lee et al. (1990) of a non-superconducting sample show a much broader (Schottky-like) feature with its maximum at about 1 K• Subsequent investigations by Allenspach et al. (1994a,b, 1995), displayed in fig. 15, revealed the consistency of the two earlier measurements under the assumption that the oxygen content of Nakazawa's sample was around 6.1 and that of Lee's sample was 6.3. The specific-heat behavior for Dy123x is similar to that observed for Nd123x in the sense that there exist sharp peaks for the samples with high and low oxygen contents, and there is an intermediate oxygen range with only short-range magnetic interactions (resulting in the broader peaks). In another sense the Dy 123x results are quite different from those of Ndl23x since there is almost no shift in the peak position and no high-temperature shoulders of the Dy123x samples.

374

RM. ALLENSPACH and M.B. MAPLE 15

. . . .

i

g oo ° oO o

. . . .

DyBa2Cu30 x

i

. . . .

i

. . . .

o o

•

go %

lO

. . . .

~

o x=6.98 [] x=6.73 o x=6.56 x=6.34 x=6.10

o~ ooo

O

o

,.~o ts'Q° a¢9

o

o

0.5

1

1.5

2

2.5

Y (K) 14

,

I ' ' '

I

. . . .

©

12

©

10 ~" ©

I

. . . .

ErBa2Cu30 x o x=7.0 (Lee) x=6.7 (Lee) x=6.4 (Lee) • x=6.3 (Simlzu) • x=6.1 (B6ttger)

Fig. 15. Oxygen dependence of the magnetic specific heat of DyBa2Cu30 x (Allenspach et al. 1994a). Superconducting samples are indicated by open symbols.

|

] I [ I I 1

8

6 "~

4

00

5.7.

0.5

1 T (K)

1.5

2

Fig. 16. Oxygen dependence of the magnetic specific heat of ErBa2Cu30 x (Lee et al. 1990, Simizu et al. 1989, B6ttger et al. 1998a). Supposedly superconducting samples are indicated by open symbols (see text).

ErBa2Cu30~

No really consistent set o f specific-heat measurements as a function o f oxygen content exist for Er123x. Simizu et al. (1989) published a set o f measurements on samples with different oxygen contents, but they obtained a discontinuous behavior below x = 6.5 which is most likely due to problems associated with the control o f the effective oxygen concentration in the sample (which was the case for most o f the samples made in the 1980s). Nevertheless, they found a broad peak in the magnetic specific heat for some o f the samples, which they fitted with a 1D XY model. Lee et al. (1990) observed a gradual decrease o f the size o f the anomaly and a broadening by going from a 90 K to a 60 K superconductor and finally to an insulating sample, similar to the behavior

HEAT CAPACITY

375

observed for Dy123x (see sect. 5.6). A recent measurement by B6ttger et al. (1998a) for a sample with x=6.10(1) seems to confirm the similarity with Dy123x: fig. 16 displays a speculative sequence of magnetic specific-heat anomalies based on an assumed consistency with Dy123x and the observations made for Er2Ba4Cu7Ol4+x (see sect. 8). Again we have sharp peaks for high and low oxygen contents, and broad peaks (shortrange correlations) in between.

6. Discussion of the experimental results of RBa2Cu3Ox

For RBa2Cu3Ox with R=Pr, Nd, Sm, Gd, Dy, Er, presented in the previous sections, the magnetic ordering for x = 7 is of antiferromagnetic and long-range character, while for oxygen concentrations close to x = 6, the specific-heat data show at least a clear tendency towards long-range magnetic ordering. For R=Nd, Sm, Dy, Er in an intermediate oxygen range (coinciding with the metal-insulator transition), only shortrange magnetic interactions were observed by specific heat and, partially, by neutron diffraction measurements (see ch. 199 of this volume). For R = P r and Gd, the number of samples investigated with different oxygen contents is too small to disprove the presence of such an oxygen regime with solely short-range magnetic ordering. Based on consistency with the other rare earths, we would expect such a regime. Due to this lack of information in the case of Pr and Gd, we will restrict our modelling of the magnetic specific heat to Nd, Sm, Dy, and Er. As a first step we will concentrate on the oxygen regime where the samples are superconducting or just becoming semiconducting (i.e., for Nd: x > 6.6; Sm: x > 6.4; Dy: x > 6.3; Er: x > 6.3). We have been successful in the past in modelling the magnetic specific heat of NdBa2Cu3Ox (Allenspach et al. 1994a,b, 1995). Since the Nd ions are almost exclusively coupled via the exchange interaction (see table 3), Nd123x is a good candidate for monitoring the variation of the exchange constants as a function of oxygen content. The magnetic specific-heat behavior over the whole range of oxygen concentration (7 > x > 6.6) could be explained assuming an increasing in-plane anisotropy of the exchange constants and a decay of the long-range magnetic ordering into magnetic clusters with antiferromagnetic coupling in one direction and ferromagnetic coupling in the other direction (fig. 17a). The exchange anisotropy was gradually changed from Ising to Heisenberg (fig. 17b) based on neutron spectroscopy investigations (Allenspach et al. 1994c). The only unrealistic result was the relatively large ferromagnetic coupling in the diagonal (via the Cu), originating from the very limited cluster size. The calculation of magnetically coupled Heisenberg clusters is very tedious and is limited to a relatively small cluster size. Hence, for the modelling of the specific heat of Sm, Dy, Er, the Ising model was used, keeping in mind its limitations and shortcomings. Nevertheless, the results seem to be quite reasonable. For the calculation, the Hamiltonian given in eq. (6) was used as explained in sect. 4.2. Since the ordered magnetic moment obtained for Sm in Smga2Cu307 (Guillaume et al. t995) is very small, the dipolar interaction can be neglected for Sm123x (as well as for Nd123x) and the direction of the magnetic

376

RM. ALLENSPACH and M.B. MAPLE

0.3

.

0.2 >

.

.

.

m

.

.

.

.

n

.

.

.

i

.

"r---J-----l----I~

N d B azC u s 0 ~

"

0.1 0.0 •

J

-o.1 -Q)°o o 0 0 0 -0.2 (a) -0.3 . . . . . . . . . . . . . . . . . . l.O .

.

.

.

i

.

.

.

.

i

.

.

.

.

i

.

.

.

.

0.14 S"

o.12

0.8

O.lO

0.6 "-~ ~

0.4

0.08 0.06 ~. 0.04 ~,

~

0.2

0.02

o.o

o.oo % 7

6.75

6.5 6.25 oxygen content x

6

Fig. 17. (a) Magnetic exchange constants and (b) exchange anisotropy ( , and dotted line) and inverse correlation length (A and solid line) obtained for NdBazCu3Q (Allenspach et al. 1995).

moment does not play a role in our calculation. For Dy and Er, a good part of the observed magnetic properties is due to the dipolar interaction and its competition with the exchange interaction. The magnetic phase diagram for an exchange-coupled and dipolar system has been published for moments parallel to the c-axis (MacIsaac et al. 1995) and perpendicular to the c-axis (MacIsaac et al. 1994). These two phase diagrams for Dy and Er, respectively, are displayed in fig. 18. Dy1237 lies clearly in the A/A regime (antiferromagnetic coupling in both plane directions). No variation in purely antiferromagnetic exchange constants will result in a phase transition into one of the other regimes or will result in a decay of the long-range into a short-range magnetically ordered system. Hence, from the observed magnetic specific-heat data and from analogy to Nd123x, we can conclude that for Dy123x at least one of the exchange constants has to become ferromagnetic upon reduction of the oxygen content. Er1237 is situated in the A/F regime (weakly coupled antiferromagnetic in the a-direction and strongly ferromagnetic in the b-direction due to the dipolar interaction). A variation of the exchange constant in either direction will finally result in a phase transition into the A/A or F/F regimes, respectively, (F/F: ferromagnetic in both directions). For both Dy and Er, an approach to a phase-transition line does not alter the character of the

HEAT CAPACITY

377

1.6

i

1.4

i

1.2

..............

(Dy) /

1.0 disordered

[,.z 0.8

]

0.6 i

0.4

!

4

ii~ 1: F

0.2 0.0 -0.05

2 ......

0

0.05 0.10 0.15 exchange constant J (meV)

i,,,

0.20

0.25

1.6

...........

1.4

i m±c' / (Er)/

1.2

i

1.0 0.8 0.6 A/A ~

0.4 0.2

, I I

(b)

,'

!

F/F

A~

I 0

I

,

'-~).06

,

,

i

,

-0.04 " '-0.()2' ' ' () ' ' '0.62' ' '0.04 exchange constant J (meV)

Fig. 18. Phase diagrams of (a) DyBa2CI130~ and (b) ErBa2Cu30 x (Maclsaac et al. 1995 and 1994, respectively). The arrows indicate the position of the samples with x 7 obtained by our dipolar and exchange interaction calculation.

magnetic specific heat (except for the lowering of the ordering temperature) despite the fact that exactly at the transition line the effective coupling in one direction has to be zero. Such a 1D coupling would result in a broad peak in the specific heat, but the range where a variation of the character of the ordering is observed is much broader in oxygen content than expected from these theoretical considerations. Hence, the neighborhood of these phase-transition lines plays an important, but not an exclusive, rote in the behavior of the magnetic ordering. A strong inhomogeneity in the electronic sublattice was observed in the oxygen range between x = 6 . 3 and x = 6 . 9 by neutron spectroscopy (Mesot et al. 1993b) and other local probes (see Vol. 30, ch. 194 of this Handbook). Such an inhomogeneous distribution of charge carriers is expected to result in a high inhomogeneity of the exchange constants. We took this effect into account by distributing the exchange constants for samples with x < 6.9 randomly on the bonds in a- and b-directions. The calculated magnetic specific-heat anomalies for Sm, Dy, and Er are displayed in fig. 19 and the exchange constants as a function of oxygen content in fig. 20. The anomalies reproduce the observed behavior quite well except that they are broader in the calculations due to the restriction to relatively small clusters (10x 10). The exchange

378

RM. ALLENSPACH and M.B. MAPLE

'

'

I

. . . .

I

. . . .

. . . .

i

10

SmBa2Cu30 x

8

--x=6.98 I ..... x=6.901 ..... x=6.75 ]

6

\

g = 0.34 I.tB (//a)

<1

2

" ~

,;I

0 0

I

I

r

i

I

0.5 . . . .

10

I

1 . . . .

(b)

I

i

,

,

,

T (K) . . . .

I

i

f'~

~.1

"7

i

,

,

I

i

i

r

2 . . . .

I

r

2.5 . . . .

DyBa2Cu30~

1--x=6.981

,I I. I ~l '~ ~ ff~ ~ /

8

i

1.5

I..... x=6.731 . . . . x=6.561 I---'x=6"341

!

-~

4

~#.¢-" o o

0 10

"7

~D

0.5 ....

1 ~ ....

1.5 T (K) ~ ....

2

2.5

~ ....

6

©

4

~.2;:~.~.

_ g = 4"9 gB (l/b)

.,q

0

0

0.5

. . . .

,

. . . .

1 T (K)

,

1.5

2

Fig. t 9. Calculated oxygen dependence of the magnetic specific heat of (a) SmBa2Cu3Oy , (b) DyBa2Cn30~, and (c) ErBazCu3Q in the superconducting regime. These figures have to be compared with figs. 13, 15, and 16, respectively.

HEAT CAPACITY

379

20

S£ai:;ol

0 -20 -40 "~

-60 -80 -100 ~

O

-120

6.8

6.6

6.4

6.2

oxygen content x 40 . . . . . . . . .

I .........

I .........

~ ~ ~uyt~azt~%0 x

I .........

.

20

~-

0 .............;,,f..~..2 ..............................................................................

g j

-20 tD~

-40 "~,,

(b)

iiiiill,l

-60

......... 6.8

I ......... 6.6

I ...... 6.4

,,,

6.2

oxygen content x 10

. . . . . . . . .

, . . . . . . . . .

, . . . . . . . . .

,.-.-~

....

ErBa2tSu3U x ~ ~ ~ 5-;:e, .................................................. 0 ................................................... -10 t

~"

-20

-30 -40

-%

-50

(c)

-60

. . . . . . . .

®I , , , , , , , , , [ ,

6.8

. . . . . . . .

I

. . . . . . . . .

6.6 6.4 oxygen content x

6.2

Fig. 20. E x c h a n g e constants obtained f r o m the calculations shown in fig. 19 for (a) S m B a a C u 3 0 ~ , (b) D y B a 2 e u 3 0 ~ , a n d (c) E r B a 2 C u 3 Q . Closed symbols indicate well-defined values in a - a n d b-direction, wbile open symbols indicate a r a n d o m distribution o f the values in the two directions. The crosses for Sm stand for a superposition o f b o t h (see text). The lines are guides to the eye.

380

PM. ALLENSPACHand M.B. MAPLE

constants behave consistently and similar to that obtained earlier for Nd. The only difference is that for Nd the cluster size and the exchange anisotropy were varied, while for Sm, Dy, and Er the cluster size was kept fixed and only Ising spins were used. Since the magnetic correlation length is reduced by the random placement of the exchange constants, this has the same effect as the reduction of the cluster size. The microscopic mechanism responsible for the rather unusual behavior of the coupling constants is unknown, but it can be argued that the larger anisotropy at x = 6.9 compared to x = 7 is due to charge ordering (e.g., striped phase) which favors a distinctly different coupling in the two plane directions. The charge distribution in samples close to x = 7 is expected to be more metal-like, and the resulting coupling constants in the plane almost equal. For superconducting samples with oxygen contents below 6.9 - according to our results - the charge-carrier density allows charge ordering only on a local scale which is independent of the crystal directions. As a second step, we will briefly compare the magnetic ordering at oxygen contents below the metal-insulator transition: As we discussed in sect. 5, the magnetic ordering temperature is almost the same for x = 6 as for x = 7 Sm, Gd, Dy, and Er compounds. According to neutron diffraction data, the magnetic ordering in samples with x = 6 tends to be more two-dimensional in character or at least a mixing of antiferromagnetic and ferromagnetic coupling along the c-direction was observed (see ch. 199 of this volume). This disruption or disorder in the magnetic interaction along c might be due to a double exchange interaction (Zener 1951, de Gennes 1960) present in these slightly doped systems. The moment direction is identical for x = 7 and x = 6 and the size of the moment is almost the same. The lowest oxygen content for reduced samples is of the order of x = 6.1, not 6. For x = 6, no double exchange would be expected and the magnetic ordering would be almost identical to the fully doped samples. In the case of Pr and Nd, a large shift in the magnetic ordering temperature (by going from x = 7 to 6) was observed from 17 to l l K in the case of Pr and 0.5 to 1.6K for Nd, respectively. These two lanthanides are among the largest atoms in the lanthanide series, and hybridization cannot be neglected. (In the case of Pr this even destroys superconductivity.) Canting of the magnetic moments relative to the c-axis - which is a feature predicted by the theory of the double exchange - was observed for Pr with x = 6 and x = 7 (Boothroyd et al. 1997) and for Nd with x=6.3 (Clinton et al. 1993). Such a canting lifts the high symmetry of the lanthanide site, and a coupling between the Cu spins (which order magnetically in the semiconducting regime) and the lanthanide spins can be realized. The hybridization might strengthen this coupling and explain why such a canting of the moments and variation of the ordering temperature was not observed for the other lanthanides. An alternative approach to understand the appearance of ferromagnetic interactions and canting of the moments is provided by the theory of Gongalves da Silva and Falicov (1972) who investigated the competition of superexchange and RKKY interaction as a function of free charge carriers. This competition resulted - in the simple model systems - in a series of different magnetic phases (antiferromagnetic, canted antiferromagnetic, helical,

HEAT CAPACITY

381

and ferromagnetic). This alone would result in a series of different long-range magnetic phases by going from x = 6 to 7, but would not explain the short-range order. Hence both the large electronic disorder in the CuO2 planes and the competition of superexchange and RKKY interaction are needed to account for the observed behavior of the magnetic lanthanide ordering in the R123x systems.

7. Experimental results for RBa2Cu408 The RBa2Cu408 (R124) compounds are close relatives of the R123 compounds. The R124 unit cell consists of two R123 units with double CuO chains which leads to a shift of the lanthanide ions in consecutive layers by half a unit cell in the a- and b-directions (see ch. 199 of this volume). This leads to frustration and almost pure 2D magnetic ordering. In contrast to the R123 compounds the oxygen content of the R124 compounds is fixed. While there are quite a few neutron diffraction papers about the magnetic ordering of the lanthanides in these compounds (see Fischer and Medarde 1998 and references therein), to our knowledge no specific-heat data have been published. This lack of data is mainly due to the fact that these compounds order magnetically in a long-range fashion only, and hence the information obtained from specific heat in addition to the neutron scattering data is very limited. Our group performed specific-heat measurements to study the magnetic ordering of Dy in a DyBa2Cu408 sample obtained from Professor E. Kaldis (ETH Zurich, Switzerland). The magnetic ordering is suppressed by an applied magnetic field as expected for antiferromagnets (fig. 21). Fitting the specific-heat curve with the 2D Ising model alone yields -60 and -22 ~teV for J~ and J2, respectively. Dipolar interactions would result in an ordering temperature of only 0.47K (~]]c, ]/~]=5.9/~B). The combination of dipolar and exchange interactions reproduces the observed ordering temperature of 1 K for J~alc =-21 ~teV (uncorrected) and Jcorr = 7.4 ~eV (corrected with the modified de Gennes factor). Jcorr for Dy124 is about 3 times larger than Jcorr for Dy1237 (table 3), indicating a larger hybridization in the former compound. This result is supported by pressure measurements (fig. 22) where the relative change of the ordering temperature under pressure (dTN/TN)/dp was found to be 0.0147Kkbar 2, which is about 3 times more than for Dy1237 (see end of sect. 3.2). The CEF splitting for Dy124 (Roessli et al. 1993b), which is about the same as that for Dy1237 (Allenspach et al. 1989), consists of a doublet ground state isolated by more than 3 meV from the next-higher doublet. Hence, the entropy change expected for the magnetic ordering is at least R In 2, but only about 80% of this value is observed (fig. 23). Under pressure, the entropy increases to nearly the expected value. This implies that under ambient pressure not all the Dy ions participate in the magnetic ordering, and that applied pressure leads to a better magnetic coupling, resulting in a higher ratio of magnetically coupled Dy ions.

382

P.M. A L L E N S P A C H and M.B. M A P L E i

.

.

.

-"" 6

.

.

.

.

.

.

i

DyBa2Cu408

I,

ooo1 0.25 T • 0.50 1.00 • 2.00

I:3OD

°

@'1,

DO

~o D • []

DO

o

D

D • D

Do

4B

3

D

~

9•

•

no

[]

•

•

•

D~•

i

i

•

i

i

i

I

i

n

•

ii

i

i

i

,

n

,

i

i

I

]

i

i

i

i

i

+

I

f

i

2

3

T (K) Fig. 21. Magnetic field dependence of the magnetic specific heat of DyBa2Cu40 s.

n

• 0 kbar[ [ .... 3 kbar]

8

~

DyBa2Cu408

6 ©

B

4 o"

"~-

,"

<1

"o

o" o..O.~ .

.e,"

o-

+

0.5

~

I

1 T (K)

i

.5

Fig. 22. Pressure dependence of the magnetic specific heat of DyBa2Cu40~.

HEAT CAPACITY 1.0

,

,

I

,

383

I

DyBa2Cu408 o

0 kbar I

0.8

3kb.r I

o

o

o

o

o

o

o

o

o

o o o o o

•

o

~"

0.6

J o

•

•

o o

•

• •

o

r.~

.,~

0.4

/

0.2

0.0

I

I

0

I

I

I

~

~

,

I

~

~

~

~

I

,

,

~

,

2 3 4 T (K) Fig. 23. Pressure dependence of the entropy of DyBa2Cu4Q. Note the increase of the entropy towards the expected value of R In 2 under hydrostaticpressure. 8.

Experimental

results

1

for

R2Ba4Cu7014+x

8. l. Introduction

Like the R123 and the R124 compounds, R2Ba4Cu7014+x (R247) contains superconducting, buckled CuO2 planes and linear C u - O chains. The structure of R247 consists of alternating 123 and 124 blocks in the c-direction. While the R 3÷ ions are surrounded by two equivalent CuO2 planes in R123, they are surrounded by two inequivalent CuO2 planes in R247. The single chains can be oxygen depleted as in the R123 phase, but the R247 compound remains superconducting due to the presence of stoichiometric R124 blocks. Removing oxygen from the single chains in the 123 blocks decreases Tc from 95 to 3 0 K (Genoud et al. 1992). In the R2Ba4Cu6+nO14+n (n=0,1,2) family, the R - R distance along the a- and b-directions is about 3.9 A, while the distance along the c-direction is three times larger. Although the large separation between the R ions along the c-axis in R124 and R123 should favour 2D ordering, we know from neutron diffraction that 3D ordering is also realized. It is therefore not surprising that Er247 exhibits 3D ordering of the Er ions as well (B6ttger et al. 1997a). Since a transition from 3D to 2D ordering of the Er ions can be achieved by oxygen removal in Er123 (Maletta et al. 1990), we expect a similar behaviour for Er247 upon oxygen depletion. An investigation of the dependence of the type of magnetic order upon x in Er2 B a4 Cu7014 + x is of special interest because Er247 is expected to be the only R247 compound that shows 3D ordering of the lanthanide ions (Zhang et al.

384

P.M. ALLENSPACH and M.B. MAPLE

1992). Thus, the presumable transition from 3D to 2D order in R247 compounds should only be observable in Er247. Correspondingly, pure 2D ordering should be observed for, e.g., Dy2Ba4CuvOis as was confirmed by neutron diffraction studies (Zhang et al. 1992). 8.2. Magnetic specific heat The oxygen dependence of the magnetic specific heat was reported for D y 2 B a 4 C u 7 0 1 4 + x (B6ttger et al. 1998b) and Er2Ba4Cu7014+x (B6ttger et al. 1997b); these data sets, together with recent measurements, are displayed in figs. 24 and 25, respectively. As for Dy123x, the anomaly in the magnetic specific heat due to the magnetic ordering remains at almost the same temperature under oxygen depletion but becomes broader and smaller. We expect that for further depletion this trend would continue towards a behavior similar to that found for Dy123x with x = 6.34 (i.e., a Schottky-like anomaly). A fit with only the anisotropic 2D Ising model (line in fig. 24) yields -71 and -17 ~teV for Jl and J2, respectively. Dipolar interaction would result in an ordering temperature of only 0.64K (/~ II c, I/zt =7/~B, according to Zhang et al. 1992). The combination of dipolar and exchange interactions reproduces the observed ordering temperature of 1.0 K for Jcalc ----12 ~teV (uncorrected) and Jcorr =--4.2 geV, which is 50% more than found for Dy1237 (see table 3). The magnetic specific heat of Er2Ba4CuTO14+x differs from that of Er123x insofar as the anomaly for x=0.3 resembles more those observed for NdBa2Cu30690 and SmBa2Cu306.90, but the tendency (long-range order for high and low oxygen contents and disorder in an intermediate range) is the same in both Er compounds. The exchange

12 .~,

•v

10

~

•

Dy2Ba4Cu7014.95

[]

Dy2Ba4Cu7014.45

3 ~10

©

6

©

4 <1

2 0

0

0.5

1

1.5

'

'

2

2.5

3

Y [K] Fig. 24. Oxygen dependence of the magnetic specific heat of Dy2Ba4CuvOt4+y (B6ttger et al. 1998b). The line is a 2D Ising calculation with the exchange values given in table 3.

HEAT CAPACITY

10

,

,

,

,

i

,

,

,

,

i

,

,

~

,

1

,

,

,

385 ,

i

,

,

,

i

,

,

,

,

o x = 0.92 I x = 0.30 I x = 0.101

Er2Ba4Cu7014+x 8

,

OO 0 0 0

"7'

0

6

0

"K "6

0

0

0

0

~o c~o

.~4

r..) <1

~e~zx~zx

0

0.5

1

1.5 W(K)

2

2.5

Aaz~zxAtx~

3

Fig. 25. Oxygen dependence of the magnetic specific heat of Er2Ba4CUTO14+~.(B6ttger et al. 1997b).

constants obtained using a pure anisotropic 2D Ising fit for Er2Ba4Cu7014.92 are given by B6ttger et al. (1997b) as J ! = ± 3 9 ~teV and J2 =q z9 ~teV (taking into account the A/F ordering observed in neutron diffraction experiments by the same group, B6ttger et al. 1997a). Dipolar interaction alone would yield an ordering temperature of 0.34K (/Z I[ b, I/z[ =3.7/ZB). The combination of dipolar and exchange interactions reproduces the observed ordering temperature of 0.54K for Jca]o=-27 ~teV (uncorrected) and Jcorr = - 4 1 ~teV. This is more than twice the value of Er1237. 8.3. Discussion A division of the values for TN and J in the theoretical phase diagram of Er123 (fig. 18b) by a factor of 1.5 (due to different lattice constants and different magnetic moments) yields the phase diagram for Er247. The area of A/F ordering (which was observed in neutron diffraction) is restricted to - 2 0 ~teV ~<J ~<+1.8 geV and 0.31 K ~< TN ~<0.42 K. The observed TN of 0.54 K would be obtained for J = -28 ~teV (A/A ordering) or +8 ~teV (F/F ordering). This contradiction between experiment and theory can only be resolved by assuming an anisotropic exchange interaction in the plane. No unambiguous set of parameters (J! , J 2 ) could be found, but typical sets are (-9.3, 0.0 ~teV) and ( - 1 5 , - 6 ~teV), which both fulfill the ordering-temperature condition. In contrast to all the other coupling constants mentioned so far, for Er247, J1 can be identified with Ja and J2 with Jb due to the moment direction and the necessity for anisotropic exchange values. At lower oxygen contents (x = 14.3), the shape of the magnetic anomaly strikingly resembles the shape of a very anisotropic 2D Ising curve (as observed for Nd123 with x = 6.90). But in contrast to the Nd compound, the size of the anomaly does not

386

P.M. A L L E N S P A C H

and M.B. MAPLE

4 Er'B'a'Cu b'4~ ' *.' . . .. .. . . 2 4 7 , .~ ,,,o,°~

(b)

' ' ' '(aiJj

. . . . . . **°%%° **o**

] "''"-~ ..........

x

:

1

o

.:

00

,

:

#°

f 00

2.

.

*°

.,"]

•

.

*

0.00 T

O.IOT

" ",.

o:2

.." , ~ . 1.00T

200T

1

2 3 4 0 1 2 3 4 T (K) T (K) Fig. 26. (a) Magnetic field dependence of the magnetic specificheat of Er2Ba4CUTOl4.3 (B6ttger et al. 1998a). (b) Calculation of the magneticfield dependenceusing a superposition of two different exchange contributions and the dipolar interaction (see text). match the theoretical curve. This may be due to the fact that the order is not really long-range, leading to a kind of smeared-out peak. But neutron-spectroscopy data (with indications for electronic cluster formation) together with the structural peculiarity of the 247 compounds (alternating 123 and 124 blocks) point more in the direction of a superposition of two individual magnetic anomalies simultaneously present in the sample. B6ttger et al. (1997b) have attributed these two contributions to areas in the samples with almost identical electronic and magnetic properties as found in the fully oxidized sample, and areas which are strongly disturbed electronically and magnetically. The inclusion of the dipolar interaction into the analysis results in an additional set of exchange constants (for the disturbed regions) which cannot be obtained unambiguously. (A set which reproduces the anomaly originating from disturbed regions is, e.g., -70/+30 gem where these two values are randomly distributed along a and b.) In order to reduce the parameter space for possible values for J~, and Jb in the undisturbed and disturbed regions, we performed specific heat measurements in applied magnetic fields for Er2Ba4Cu7014.3 (fig. 26a). From this figure one can see that in fields above 0.5 tesla the magnetic specific heat manifests itself as a Schottky anomaly due to the splitting of the ground-state doublet [with a typical g-value of 6.8 which is close to the value 7.1 expected in a powder sample from gx = 5.7, gy = 9.7, and gz = 4.8, calculated by B6ttger (1998) from her neutron spectroscopy results]. Hence, magnetic interactions are effective only in a regime below 0.5 tesla. A reasonable (double) set of exchange parameters has to fulfill these criteria. Since Er247 is far from being an ideal Ising system (as can be seen from the g-values), more than qualitative agreement is not expected from such an analysis. The magnetic-field dependence of the magnetic specific heat of an Ising system [with g = 9.7; ./a = - 9 . 3 ~eV and .l b = 0 ~teV (undisturbed regions) and Ja/b = - - 7 0 / + 3 0 geV (disturbed

HEAT CAPACITY

387

regions), respectively] is displayed in fig. 26b. As can be seen the qualitative behavior of the measured and the calculated curves is similar, with the increase of signal at a field of 0.1 tesla (in the calculation) being the largest discrepancy. This increase is due to an insufficient stiffness of the magnetic ordering in the undisturbed regions in the calculation. In addition, the calculation does not present a powder average. Such an average would diminish the 0.10-tesla increase and would restrict the shift of the Schottky anomaly to the observed value (i.e., a shift corresponding to g ~ 7).

9. Conclusions We have presented above a phenomenological approach to the available magnetic heatcapacity data for the 123, 124 and 247 compounds. The model includes dipolar and exchange interactions. At high oxygen contents, all samples which order magnetically do so in a long-range fashion. Oxygen reduction leads in all these compounds to an increase of the anisotropy of the in-plane magnetic coupling due to the competition between the superexchange and the RKKY interactions and - for a restricted oxygen regime - to strong disorder in the electronic sublattice. Finally, at low oxygen contents, long-range magnetic ordering is recovered due to the reappearance of structural and electronic order in the totally reduced samples. Due to the crude theoretical tools at hand and the severe simplifications made (e.g., Ising model for all the rare earths), we did not actually fit theoretical results to the experimental data but just presented theoretical curves which demonstrate the qualitative agreement. The aim of this contribution should be to encourage more elaborate theoretical studies of the magnetic interactions and their dependence on doping in these fascinating compounds. Besides the question concerning the origin of the observed magnetic ordering effects, there are still open questions and unexplored effects on the experimental side: - No experimental data exist for Pr123 in the intermediate oxygen-content regime. - In the intermediate oxygen regime of Sm123, the center of mass of the magnetic anomaly is higher than that of both the fully oxidized and the fully reduced samples. Is the transition towards the reduced sample sharp or gradual as towards the oxidized sample? - Does an oxygen regime also exist for Gd123 where only short-range ordering is present? What is the origin of the 1 K shoulder? - The specific-heat data of Er123 originate from different sources and the samples were not consistently characterized. - The evolution of the magnetic ordering in Er247 is not yet studied sufficiently well and, hence, the origin of the formation of the two different magnetic regimes is still unclear.

Acknowledgment Financial support by the US Department of Energy under Grant No. DE-FG03-86ER45230 and the Swiss National Science Foundation is kindly acknowledged.

388

RM. ALLENSPACH and M.B. MAPLE

References Alekseevskii, N.E., A.V. Gusev, G.G. Devyatylda, A.V. Kabanov, VN. Narozhnyi, VI. Nizhankovskii and E.R Khlybov, 1988, JETP Lett. 48(10), 585. Allenspach, E, A. Furrer and E Hulliger, 1989, Phys. Rev. B 39(4), 2226. Allenspach, E, B.W. Lee, D.A. Gajewski and M.B. Maple, 1993, J. Appl. Phys. 73(10), 6317. Allenspach, R, M.B. Maple and A. Furrer, 1994a, J. Alloys & Compounds 207-208, 213. Allenspach, E, B.W. Lee, D.A. Gajewski and M.B. Maple, 1994b, Physica C 235-240, 1753. Allenspach, E, J. Mesot, U. Staub, M. Guillaume, A. Furrer, S.-L Yoo, M.J. Kramer, R.W. McCallum, It. Maletta, H. Blank, H. Mutka, R. Osborn, M. Arai, Z. Bowden and A.D. Taylor, 1994e, Z. Phys. B 95, 301. Allenspach, P., B.W Lee, D.A. Gajewski, VB. Barbeta, M.B. Maple, G. Nieva, S.-I. Yoo, M.J. Kramer, R.W McCallum and L. Ben-Dor, 1995, Z. Phys. B 96, 455. Allenspach, E, M.C. de Andrade and M.B. Maple, 1996, Czech. J. Phys. 46(Suppl. $4), 2103. Alloul, H., E Mendels, G. Collhr and E Monod, 1988, Phys. Rev. Lett. 61, 746. Atake, T., 1991, Thermochim. Acta 174, 281. Basu, B.K., S.M. Pattalwar, R.N. Dixit, S.Y. Shete, G. Chandra, A.K. Nigam and R.S. Sannabhadti, 1989, Physica C 159, 379. Basu, B.K., S.M. Pattalwar, R.N. Dixit and S.Y. Shete, 1991, Bull. Mater. Sci. 14(3), 655. Bloom, S.H., M.V Kuric, R.R Guertin, C.-S. Jee, D. Nichols, E. Kaczanowicz, J.E. Crow, G. Myer and R.E. Salomon, 1987, J. Magn. Magn. Mater. 68, L135. Boothroyd, A.T., L. Longmore, N.H. Andersen, E. Brecht and Th. Wolf, 1997, Phys. Rev. Lett. 78(1), 130. B6ttger, G., 1998, Ph.D. Thesis (ETH Zttrich). B6ttger, G., R Fischer, A. D6nni, R Berastegui, Y. Aoki, H. Sato and E Fauth, 1997a, Phys. Rev. B 55, R12005. B6ttger, G., R Allenspach, A. D6nni, Y. Aoki and H. Sato, 1997b, Z. Phys. B 104, 195. B6ttger, G., R Allenspach, A. D6nni, Y. Aoki, T. Namiki and H. Sato, 1998a, unpublished. B6ttger, G., R Allenspach, J. Mesot, A. D61mi, Y. Aoki and H. Sato, 1998b, J. Alloys & Compounds 275 277, 560.

Brown, S.E., J.D. Thompson, J.O. Willis, R.M. Aikin, E. Zirngiebl, J.L. Smith, Z. Fisk and R.B. Schwarz, 1987, Phys. Rev. B 36(4), 2298. Buchgeister, M., W Hiller, S.M. Hosseini, K. Kopitzki and D. Wagener, 1990, in: Proc. Int. Conf. on Transport Properties of Superconductors, Rio de Janeiro, Brazil (1990), ed. R. Niclosky (World Scientific, Singapore) p. 511. Causa, M.T., S.M. Dutrus, C. Fainstein, G. Nieva, H.R. Salva, R. Sanchez, L.B. Steren, M. Tovar and R. Zysler, 1987, Int. J. Mod. Phys. B 1(3&4), 989. Causa, M.T., C. Fainstein, G. Nieva, R. Sanchez, L.B. Steren, M. Tovar and R. Zysler, 1988, Phys. Rev. B 38(1), 257. Chattopadhyay, T., H. Maletta, W. Wirges, K. Fischer and EJ. Brown, 1988, Phys. Rev. B 3g(1), 838. Chmist, J., A. Lewicki, Z. Tarnawski, A. Kozlowski, J. Zukrowski, W.M. Woch, A. Kolodziejczyk and K. Krop, 1988, Aeta Phys. Pol. A 74, 757. Clinton, T.W, J.W. Lynn, B.W. Lee, M. Buchgeister and M.B. Maple, 1993, J. Appl. Phys. 73, 6320. de Gennes, RG., 1960, Phys. Rev. 118(1), 141. de Jongh, L.J., and A.R. Miedema, 1974, Adv. Phys. 23(1), 1~60. Dirken, M.W, and LJ. de Jongh, 1987, Solid State Commun. 64(9), 1201. Dr6ssler, H., H.-D. Jostarndt, J. Harnischmacher, J. Kalenborn, U. Walter, A. Severing, W. Schlabitz and E. Holland-Moritz, 1996, Z. Phys B 100, 1. Dunlap, B.D., M. Slaski, D.G. Hinks, L. Soderholm, M.A. Beno, K. Zhang, C. Segre, G.W Crabtree, W.K. Kwok, S.K. Malik, I.K. Schuller, J.D. Jorgensen and Z. Sungaila, 1987, J. Magn. Magn. Mater. 68, L139. Dunlap, B.D., M. Slaski, Z. Sungaila, D.G. Hinks, K. Zhang, C. Segre, S.K. Malik and E.E. Alp, 1988, Phys. Rev. B 37(1), 592. Escribe-Filippini, C., K. Konate, R. Buder, J. Marcus and C. Schtenker, 1988, Physica C 153 155, 1030. Ferreira, J.M., B.W Lee, Y. Dalichaouch, M.S. Torikachvili, K.N. Yang and M.B. Maple, 1988, Phys. Rev. B 37(4), 1580. Fischer, E, and M. Medarde, 1998, in: Neutron Scattering in Layered Copper-Oxide Superconductors, ed. A. Furrer, Vol. 20 of Physics and Chemistry of Materials with Low-dimensional Structure (Kluwer Academic Publishers, Dordrecht) pp. 261-301. Fischer, R, B. Schmid, E Briiesch, E Smcki and R Unternfihrer, 1989, Z. Phys. B 74, 183.

HEAT CAPACITY Genoud, J.-Y., T. Graf, G. Triscone, A. Jnnod and J. Muller, 1992, Physica C 192, 137. Gering, E., B. Renker, E Gompf, D. Ewert, H. Schmidt, R. Arens, M. Bonnet and A. Dianoux, 1988, Physica C 153 155, 184. Gongalves da Silva, C.E.T., and L.M. Falicov, t972, J. Phys. C 5, 63. Guillaume, M., E Fischer, B. Roessli, E Allenspach and V Trunov, 1994a, Physica C 235 240, 1637. Guillaume, M., U. Staub, E Fanth, J. Mesot, A. Furrer and C. Carlile, 1994b, Physica C 223, 333. Guillaume, M., W Henggeler, A. Furrer, R.S. Eccleston and V. Trounov, 1995, Phys. Rev. Lett. 74(17), 3423. Gurevich, A.M., VI. Dotsenko, VN. Eropkin, Y.V Telezhenko and N.M. Chaikovskaya, 1990, Soy. J. Low Temp. Phys. 16(1), 27. Heremans, J., D.T. Moretli, G.W. Smith and S.C. Strite III, 1988, Phys. Rev. B 37(4), 1604. Hilscher, G., T. Holubar, G. Schaudy, J. Dumschat, M. Strecker, G. Wortmarm, Z.X. Wang, B. Hellebrand and D. Bfiuerle, 1994, Phys. Rev. B 49, 535. Ho, J.C., EH. Hot, R.L. Meng, C.W. Chu and C.Y. Huang, 1987, Solid State Commun. 63(8), 711. Jana, EC., and R.C. Bhuniya, 1993, Solid State Commnn. 87(8), 731. Jensen, J., and A.R. Mackintosh, 1991, Rare Earth Magnetism (Clarendon Press, Oxford). Jnnod, A., 1996, Specific heat of high temperature superconductors in high magnetic fields, in: Studies of High Temperature Superconductors, Vol. 19, ed A.V Narlikar (Nova Science Publisher Inc., New York) p. 1. Kadowaki, K., H.P. van der Meulen, J.C.E Klaasse, M. van Sprang, J.Q.A. Koster, L.W Roeland, ER. de Boer, Y.K. Huang, A.A. Menovsky and J.J.M. Franse, 1987, Physica 145B, 260. Kierspel, H., E Oster, H. Dr6ssler, U. Callies, H. General, H. Geus, G. Jackel, H.-D. Jostarndt, J. Langen, A. Waldorfel, R. Mtiller, W. Schlabitz and D. Wohlleben, 1994, Low temperature specific heat and AC-susceptibility of REBa2Cu307 ,~ (RE = Y, Nd, Eu, Gd, Dy, Ho, Er, Tm and Yb), in: High Temperature Superconductors: Physical and Material Properties, eds S.K. Malik and S.S. Jhah (Nova Science Publisher Inc., New York). Kobayashi, T., K. Amaya, T. Kohara, K. Ueda, Y. Kohori, Y. Oda and K. Asayama, 1987, J. Phys. Soc. Jpn. 56(11), 3805. Kohara, T., Y. Oda, Y. Kohori, I. Nakada, Y. Yamada, K. Amaya, T. Kobayashi, E. Sakagami, T. Kaneko,

389

H. Fujita, H. Toyoda and K. Asayama, 1987, Jpn. J. Appl. Phys. 26(10), L1672. Lazarev, V.B., I.S. Shaplygin, K.S. Gavrichev, LA. Konovalova, VE. Gorbunov and E.A. Tistchenko, 1988, Physica C 153-155, 1024. Lee, B.W., J.M. Ferreira, Y. Dalichaouch, M.S. Torikachvili, K.N. Yang and M.B. Maple, 1988, Phys. Rev. B 37(4), 2368. Lee, B.W., J.M. Ferreira, S. Ghamaty, K.N. Yang and M.B. Maple, 1990, in: Oxygen Disorder Effects in High-Tc Superconductors, eds J.L. Moran-Lopez and I.K. Schuller (Plenum, New York) p. 151. Maclsaac, A.B., J.R Whitehead, K. De'Bell and K. Sowmya Narayanan, 1992, Phys. Rev. B 46(10), 6387. Maclsaac, A.B., J.R Whitehead, M.C. Robinson and K. De'Bell, 1994, Physica B 194 196, 223. Maclsaac, A.B., J.P. Whitehead, M.C. Robinson and K. De'Bell, 1995, Phys. Rev. B 51(22), 16033. Matetta, H., E. P/3rschke, T. Chattopadhyay and EJ. Brown, 1990, Physica C 166, 9. Maple, M.B., J.M. Ferreira, R.R. Hake, B.W. Lee, J.J. Neumeier, C.L. Seaman, K.N. Yang and H. Zhou, 1989, J. Less-Common Met. 149, 405. Mesot, J., P. Allenspach, U. Staub, A. Furrer, H. Mutka, R. Osborn and A. Taylor, 1993a, Phys. Rev. B 47(10), 6027. Mesot, J., P. Allenspach, U. Staub, A. Furrer and H. Mutka, 1993b, Phys. Rev. Lett. 70(6), 865. Meyer, C., H.-J. Bornemann, H. Schmidt, R. Ahrens, D. Ewert, B. Renker and G. Czjzek, 1987, J. Phys. F 17, L345. Mori, K., K. Noto, M. Sasakawa, Y. Isikawa, K. Sato, N. Kobayashi and Y. Muto, 1988, Physica C 153-155, 1515. Nakazawa, Y., M. Ishikawa and T. Takabatake, 1987, Physica B 148, 404. Naumov, VN., G.I. Frolova, E.B. Amitin, V.E. Fedorov and EE Samoilov, 1996, Physica C 262, 143. Niraimathi, A.M., E. Gmelin and G. Rangarajan, 1994, Physica C 235-240, 857. Niraimathi, A.M., G. Rangarajan, R.K. Kremer and E. Gmelin, 1995a, Ceramics Int. 21, 367. Niraimathi, A.M., E. Gmelin and G. Rangarajan, i995b, Phys. Rev. B 51(13), 8503. Onsager, L., 1944, Phys. Rev. 65, 117. Oota, A., and T. Kojima, 1988, Jpn. J. Appl. Phys. 27(3), L333. Oota, A., T. Kojima and Y. Sasaki, t989, Proc. MRS Int. Meetings on Adv. Mater. 6, 341.

390

EM. ALLENSPACH and M.B. MAPLE

Plackowski, T., and D. Wlosewicz, 1996, Solid State Conmaun. 98(1), 31. Ramirez, A.R, L.E Schneemeyer and J.V. Waszczak, 1987, Phys. Rev. B 36(13), 7145. Ravin&an, N., T. Szxkar, S. Uma, G. Rangarajan and V Sankaranarayanan, 1995, Phys. Rev. B 52, 7656. Reeves, M.E., D.S. Citrin, B.G. Pazol, T.A. Friedmann and D.M. Ginsberg, 1987, Phys. Rev. B 36(13), 6915. Roessli, B., R Fischer, U. Stanb, M. Zolliker and A. Furrer, 1993a, Europhys. Lett. 23(7), 511. Roessli, B., E Fischer, M. Zolliker, P. Allenspach, J. Mesot, U. Staub, A. Furrer, E. Kaldis, B. Bucher, J. Karpinski, E. Jilek and H. Mutka, 1993b, Z. Phys. B 91, 149. Schaudy, G., T. Holubar, N. PiUmayr, G. Leitenbauer, G. Hilscher, P. Rogl and A. Hanninger, 1992, J. Magn. Magn. Mater. 104 107, 477. Schaudy, G., T. Holubar, M. Forsthuber, G. Hilscher, E. Holland-Moritz, V Nekvasil and R Rogl, 1993, in: Electronic Properties of High-T0 Superconductors, eds H. Kuzmany, M. Mehring and J. Fink, Vol. 113 of Springer Series in Solid-State Sciences (Springer, Berlin) p. 113. Simizu, S., S.A. Friedberg, E.A. Hayri and M. Greenblatt, 1987a, Jpn. J. Appl. Phys. 26(Suppl. 26-3), 2121. Simizu, S., S.A. Friedberg, E.A. Hayri and M. Greenblatt, 1987b, Phys. Rev. B 36, 7129. Simizu, S., G.H. Bellesis, J. Lukin, S.A. Friedberg, H.S. Lessure, S.M. Fine and M. Greenblatt, 1989, Phys. Rev. B 39(13), 9099. Slaski, M., B.D. Dunlap, M.V Nevitt and Z. Sungaila, 1988, in: High-To Superconductors, ed. H. Weber (Plenum, New York) p. 89. Smit, H.H.A., M.W Dirken, R.C. Thiet and L.J. de Jongh, 1987, Solid State Commun. 64(5), 695. Staub, U., E Fauth, M. Guillaume, J. Mesot, A. Furrer, R Dosanjh and H. Zhou, 1993, Europhys. Lett. 21(8), 845.

Thuy, N.E, N.T. Hien, T.D. Hien and J.J.M. Franse, 1992, J. Magn. Magn. Mater. 104-107, 489. Uma, S., W Schnelle, E. Gmelin, G. Rangarajan, A. Erb, E. Walker and R. Fliikiger, 1996, Czech. J. Phys. 46(Suppl. $5), 2677. van den Berg, J., CJ. van der Beek, EH. Kes, J.A. Mydosh, G.J. Nieuwenhuys and L.J. de Jongh, 1987, Solid State Commtm. 64(5), 699. van der Meulen, H.P., J.J.M. Franse, Z. Tarnawski, K. Kadowaki, J.C.P. Klaasse and A.A. Menovsky, 1988a, Physica C 152, 65. van der Meulen, H.P., J.J.M. Franse, Z. Tarnawski, K. Kadowaki, J.C.E Klaasse and A.A. Menovsky, 1988b, Physica C 153 155, 1018. Wang, K., Z. Chen, J. Xia, Z. Chen and Q. Zhang, 1989, Solid State Commun. 69(12), 1145. Willekers, R.W, H.C. Meijer, G. Rietfeld, E Mathu and H. Postma, 1990a, Physica B 165-166, 1333. Willekers, R.W., H.C. Meijer, H. Postma and M.J.E Menken, 1990b, Physica C 168, 99. Willekers, R.W, H.C. Meijer, E Mathu and H. Postma, 1991, Cryogenics 31, 168. Willis, J.O., Z. Fisk, J.D. Thompson, S.-W Cheong, R.M. Aikin, J.L. Smith and E. Zirngiebl, 1987, J. Magn. Magn. Mater. 67, L139. Yang, K.N., J.M. Ferreira, B.W. Lee, M.B. Maple, W.H. Li, J.W Lynn and R.W Erwin, 1989, Phys. Rev. B 40(16), 10963. Yeomans, J.M., 1992, Statistical Mechanics of Phase Transformations (Clarendon Press, Oxford). Zener, C., 1951, Phys. Rev. 82(3), 403. Zhang, H., J.W. Lynn and D.E. Morris, 1992, Phys. Rev. B 45, 10022. Zhu, C., Zhao Guo-meng and Q.-Z. Ran, 1988, Solid State Commtm. 66(7), 719. Zou, Z., J. Ye, K. Oka and 5(. Nishihara, 1998, Phys. Rev. Lett. 80, 174.

Handbook on the Physics and Chemistry of Rare Earths Vol. 31 edited by K.A. Gschneidner, Jr., L. Eyring and M.B. Maple © 2001 Elsevier Science B. F? All rights reserved

Chapter 201 A N G L E - R E S O L V E D P H O T O E M I S S I O N S T U D I E S OF U N T W I N N E D Y T T R I U M B A R I U M C O P P E R OXIDE Matthias SCHABEL

Remote Sensing Systems, 438 First St. Suite 200, Santa Rosa, CA 95401, USA

Z.-X. S H E N Department o f Applied Physics, Stanford University, StanJbrd, CA, USA

Contents Abbreviations 1. Introduction 1.1. Electronic structure 2. Photoelectron spectroscopy 2.1. Angle-resolved photoelectron spectroscopy 2.2. Resonance photoemission 2.3. Structure of Yt23 2.4. Surface termination of Y123 3. Photoemission studies of the cuprates 3.1. Normal-state electronic structure

391 392 394 396

3.2. Surface chain feature 3.3. Band structure 4. Fermi surface 4.1. Methods for Fermi surface determination with ARPES 4.2. The Fermi surface of Y123 4.3. Shadow bands in Y123 4.4. Evidence for bilayer splitting 5. Superconducting-state electronic structm'e 6. Summary References

398 399 400 403 404 406

Abbrevi~ions ARPES

Angle-resolved photoelectron spectroscopy

MIT

Metal insulator transition

BCS

Bardee~Cooper-Schrieffer

PES

Photoelectron spectroscopy

Bi2212 BZ

Bi2 Sr2CaCu2Os Brillouin zone

RESPES Resonance photoelectron spectroscopy Surface chain feature SCF

CDW DOS

Charge density wave Density of states

STM TMO

Scanning tunneling microscopy Transition metal oxide

EDC

Energy distribution curve

UHB

Upper hubbard band

EELS

Electron energy loss spectroscopy

UHV

EVHS FS

Extended Van Hove singularity Fermi surface

UPS XAS

Ultrahigh vacuum Ultraviolet photoelectron spectroscopy

HTS LDA

High-temperature superconductor Local density approximation

XPS Y123

LHB

Lower Hubbard band Linearized muffin tin orbital

Y124

LMTO

391

X-ray absorption spectroscopy X-ray photoelectron spectroscopy YBa~Cu307 YBa2Cu408

409 413 417 419 421 424 425 426 431 432

392

M. SCHABELand Z.-X. SHEN

1. Introduction

in 1986 the world of condensed matter physics was set abuzz by Bednorz and Mfiller's discovery of superconductivity in La2 xBaxCuO4 (La214) at the unprecedented temperature of 38 K (Bednorz and Mtiller 1986). Perhaps even more dramatic than the extremely high superconducting transition temperature was its observation in a highly anisotropic, conductive ceramic oxide. This discovery immediately instigated an unparalleled effort in materials synthesis, leading to dramatic increases in transition temperatures in YBa2Cu307_. ~ (Y123), Bi2Sr2CaCu208 + ~ (Bi2212), T12Ba2Ca2Cu3010 + ~ (T12223), and HgBazCa2Cu3010 + ~ (Hg1223), with the three-plane Hg1223 material having the highest transition temperature known to date, reaching over 150 K under pressure (Chu et al. 1993, Maeda et al. 1988, Schilling et al. 1993, Sheng and Hermann 1988, Wu et al. 1987). Rapid strides in crystal growth techniques were paralleled by equally rapid developments in methods of experimental characterization. Nevertheless, after more than a decade of intense study, the cuprate perovskites continue to challenge our understanding of electronic structure in general, and superconductivity in particular. Prior to the discovery of these novel materials, the Bardeen-Cooper-Schrieffer (BCS) theory provided a coherent and quantitative picture of the microscopic mechanism of superconductivity through electron-phonon coupling, although lacking in predictive power at the materials level (Bardeen et al. 1957). The astronomically high transition temperatures in oxide superconductors coupled with their other unconventional properties have necessitated a re-examination of the conventional theory, and have turned attention to other possible pairing mechanisms, particularly magnetic ones (Anderson 1987, Dagotto 1994, Dagotto et al. 1994, Fukuyama 1988, Kampf and Schrieffer 1990, Laughlin 1988, Markiewicz 1991, Millis et al. 1990, Monthoux et al. 1991, Nagaosa and Lee 1991, Scalapino et al. 1986, Schrieffer et al. 1988, Virosztek and Ruvalds 1990). Furthermore, the poorly understood aspects of electronic structure in correlated systems in general have been thrust to the forefront of theoretical efforts. Of the many experimental techniques used in the pursuit of insight into the cuprates, angle-resolved photoemission has taken a central role by virtue of its ability to directly probe both energy and momentum scales relevant to these strongly correlated materials. While each family of cuprate superconductors differs in details, all share the same fundamental structure: one or more planar CuO2 layers in which square plaquettes of Cu atoms fourfold coordinate to O atoms lying between them (fig. 1), with a comparatively inert interlayer separating the copper oxide planes. This interlayer serves as a charge reservoir for dopants, as well as stabilizing the CuO2 structure. Their highly anisotropic two-dimensional character was implicated from the outset in their anomalous properties, but the exact role of reduced dimensionality remains unclear at this time (Anderson 1990, Varma et al. 1989). Within a family, monotonically increasing critical temperatures are observed with increasing number of CuO2 units per elementary cell, from unilayers to trilayers, another behavior which remains poorly understood. Even the phenomenology of Tc dependence on the interlayer properties lacks systematics, although arguments have

ANGLE-RESOLVEDPHOTOEMISSIONSTUDIES OF UNTWlNNEDYBCO

393

@ Cu

O

O Fig. 1. The atomicstructure of the ubiquitous CuO2 planes, which are believed to play a pivotal role in the unusual properties of the cuprates. The inplane lattice constant, a, is approximately3.85A. A sketch of the Cu 3dy2 y2 and O 2px.y orbitals which hybridize to form the low energy states near the Fermi surface is superimposed.

been made for a correlation with in-plane lattice constant (Fisk and Sarrao 1995), or interlayer tunneling properties (Anderson 1990). In the superconducting state, the cuprates also manifest a number of anomalous properties. Unlike conventional superconductors, which have superconducting coherence lengths, ~, on the order of 103-104 A, the Cooper pairs in high-temperature superconductors (HTS) are localized over a region 12 15A in the plane. Parallel to the c-axis, is even smaller than the corresponding unit cell dimension (~2-5 A). This tendency to strongly localize in real, rather than momentum space may be an indicator that the fundamental interactions responsible for pairing in the cuprates are strong, short-ranged forces rather than the weaker phonon-mediated mean-field forces causing pairing in conventional BCS materials (Scalapino 1995). Alternatively, localized polaron formation may lead to significant deviations from conventional behavior (Alexandrov and Mott 1995, Salje et al. 1995). Furthermore, it has been established with a high degree of certainty in the past few years that the symmetry of the superconducting order parameter has an anisotropic form, with a ~/2 phase change associated with a 90 ° rotation along the C u - O bonds, and lines of nodes (zero order parameter) along the plane diagonal (Ott and Brawner 1995, Tsuei et al. 1995, van Harlingen 1995, Wollman et al. 1995). This sort of behavior is consistent with a magnetic pairing force based on the strong tendency for antiferromagnetic alignment of unpaired spins in these materials (Bulut and Scalapino 1995, Scalapino 1995, 1996). Stimulated by the unusual properties of the cuprate superconductors, interest in the electronic structure of highly correlated materials, particularly transition metal oxides (TMO), has experienced a renascence. These are systems for which conventional one-electron theory fails qualitatively, often predicting metallic behavior for compounds known to be large-gap insulators. In light of the great strides which have been made in understanding electronic structure from the perspective of band theory, this failure is particularly intriguing, and appears to have far-reaching ramifications for the theory of many-particle systems (Zaanen et al. 1985). In the HTS, the CuO2 planes which are responsible for carrier transport, and are believed to be the critical element for high-To behavior, clearly fall into this class, and manifest many of the hallmarks of

394

M. SCHABELand Z.-X. SHEN

strong correlations. Hybridization of the narrow band of localized Cu 3d states with an itinerant O 2p band leads to an antiferromagnetic parent compound which is a charge transfer insulator, and which rapidly evolves into a "strange metal" when doped with small concentrations of holes (Allen 1985, Fuggle and Inglesfield 1992, Zaanen et al. 1985, Zaanen and Sawatzky 1990). The curious and significant coexistence in the HTS of numerous phases with unusual magnetic order in a small doping regime parallels the empirical observation in normal superconductors that higher critical temperatures are typically associated with structural or electronic instabilities. The armaments of a vast number of experimental groups have been applied to studies of TMO in general, and HTS in particular, providing a wealth of often confusing, sometimes contradictory results (Cox 1992, Tokura and Fujimori 1995). The fact that their baffling properties are only slowly being unraveled is a testament to the complexity of these highly unconventional oxides. From the outset photoemission, particularly the angle-resolved technique (ARPES), has made critical contributions to these investigations. With improvements in materials quality and instrumentation, it has found itself at the forefront of debate centered on the electronic structure and nature of the pairing state. The lamellar nature of these materials allows, in the ideal case, unambiguous determination of the planar band structure due to the lack of dispersion between planes (Olson et al. 1989a,b). Issues of recent interest or current concern include the presence and relevance of an extended flat band van Hove singularity, existence, volume, and topology of the Fermi surface, the anisotropy of the superconducting gap and the symmetry of the order parameter, the doping dependence of the electronic structure, the appearance of a pseudogap above above the critical temperature, including the possibility of pairing without phase coherence, and the nature of the fundamental low-energy excitations (Levi 1993). 1.1. Electronic structure

An extremely simple way of visualizing the Mott metal-insulator transition is shown in fig. 2. If we consider a lattice of alkali metal atoms (each having a half-filled outer s orbital) in the metallic state (fig. 2a), it is clear that the valence electrons can decrease their kinetic energy by becoming highly delocalized, resulting in a nearly uniform probability amplitude throughout the crystal. Because the strongly screened Coulomb interaction with other electrons is weak and short-ranged, this delocalization is favored. However, as we adiabatically increase the interatomic distance, the orbital overlap, and consequently, the bandwidth, W, diminish rapidly. Ultimately we arrive at the situation depicted in fig. 2b, in which there is a lattice of isolated atoms, each independent of the others. In this limit of very large lattice parameter, conduction must proceed via hopping of a single electron between sites. Each atom is essentially unaffected by the presence of its neighbors due to the exponentially shrinking inter-site matrix element, and the formal imposition of Bloch's theorem is meaningless. Hopping of parallel spins is absolutely prohibited by Pauli exclusion in the case of a single orbital, while antiparallel spins may doubly occupy the orbital, in the process incurring an energy cost, U, due to the mutual Coulomb repulsion of two electrons on a single site.

ANGLE-RESOLVED PHOTOEMISSION STUDIES OF UNTWlNNED YBCO

W

395

t,

®®®®®®® EF ~----U

®

®

® E(s 1)

,c, A A A A A

E(s 2) ~--- Ueff ---~

®®®®®

LHB

UHB

Fig. 2. Schematic illustration of the metaHnsulator transition and formation of Hubbard bands in narrowbandwidth materials. In the itinerant metal (a), strong overlap of the valence orbitals leads to formation of a broad hybridized energy band of width W. Electrons are essentially completely delocalized to minimize their kinetic energy. At the opposite extreme (b) is the atomic case, where the eigenstates arc well defined and the Coulombic repulsion between two etectTons in a single orbital costs an energy U. Transition-metal oxides and similar materials exist in the intermediate regime suggested in (c), with an effective Coulomb repulsion, larger than their bandwidth: Uea->> W/2, separating the upper (UHB) and lower (LHB) Hubbard bands.

The fascinating intermediate regime lying between itinerant electrons and isolated atomic states is illustrated in fig. 2c. Clearly, the extreme cases o f metallic and atomic behavior must be separated by a phase transition demarcating the boundary between these qualitatively different situations, from a single strongly hybridized band o f width W to two distinct energy levels separated by a finite U = E(s 2) - E(st). It is precisely this phase boundary which defines the metal-insulator transition (MIT) in the M o t t - H u b b a r d model, and is crucial for understanding the properties o f TMO. The example constructed above is an artificial one, and neglects a host o f other competing phenomena which can arise, including small polarons and Jahn-Teller distortions, charge density waves (CDW), local phase separation in the form o f spin, charge, or orbital ordering, and structural phase transitions (such as the Peierls transition) in which changes in local or long-range order result in a lower ground-state energy. It is the fortuitous chemical conjunction o f oxygen, which readily hybridizes to form delocalized ligand band states, and a 3d transition metal, having localized 3d and itinerant valence states, which gives the transition metal oxides their unique properties. By forming structurally stable crystals in which the O atoms act as spacers forcing the transition-metal atoms apart to reduce 3d orbital overlap, these systems exist in the parameter regime where the M o t t - H u b b a r d transition occurs, and reveal a wealth o f fascinating properties. Members run the gamut from conventional metals, semiconductors, and insulators to materials exhibiting metal-insulator transitions, a class which includes the cuprates. The effective Coulomb repulsion, Ueft', represents the energetic penalty incurred when the 3d orbital is doubly occupied, and is reduced from

396

M. SCHABEL and Z.-X. SHEN

that in the bare atom by screening effects. Nevertheless, it is clear that, when Ueff >> W/2, we are in a regime qualitatively different from the picture described by band theory.

2. Photoelectron spectroscopy The simplest gedanken photoemission experiment (fig. 3) consists of monochromatic ultraviolet (UPS) or X-ray (XPS) photons from a synchrotron, X-ray anode, or glow discharge lamp impinging on the surface of a solid sample, where a fraction are absorbed by electrons lying in bound states. Electrons bound by less than h v - ~, where q~ is the sample work function, escape the sample and propagate to a detector where their kinetic energy distribution is measured. Conservation of energy requires that the difference in energies between the initial state, with a photon in free space and an unperturbed solid, and the final state, with a photoelectron in free space and a single hole propagating in the solid, equal the energy of the photon. This does not, however, preclude the possibility of interactions between the photoelectron and photohole; much of the difficulty in interpreting photoemission data arises from such "final-state effects" (Cardona and Ley 1978a,b, Htifner 1996). In addition, the extremely short mean free path for inelastic scattering of photoelectrons, typically on the order of 5 A at the photon

CONDUCTION BAND

~ 1) EXCITATION

VACENOE BA"D 2) TRANSPORT

j 3) ESCAPE

Fig. 3. A schematic depiction of the photoemission process within the context of the three-step model. An incident photon, with energy h% is absorbed by a bound electron ( e ) in the solid, which is subsequently photoemitted with energy h v - q~, leaving behind a positively charged hole (h~)~

ANGLE-RESOLVED PHOTOEMISSION STUDIES OF UNTW1NNED YBCO

397

YBa2Cu3OT_ valence band

@ 4iY 4"

f~

@4

d

1.5.oo

~o.oo

5.00

o.oo

Binding energy (eV) Fig. 4. The photoemission process, showing the valence band of Y123, along with the Ba 5p core level, and the background of inelastically scattered electrons. The incident photon is absorbed in the solid, producing a photoexcited electron and a positively charged hole. Energy losses during transport of the outgoing photoelectron give rise to the background, which should be proportional to the integrated weight of photoelectrons at higher kinetic energies. Both unscattered and scattered photoelectrons experience a step loss in energy passing through the surface work-function barrier (~) before propagating to the detector.

energies used in UPS (slightly longer in XPS), (Schabel et al. 1991, Seah and Dench 1979) makes the technique extremely sensitive to surface properties, necessitating the maintenance of ultrahigh vacuum (UHV) conditions (~ 10 11torr), along with scrupulous care in sample preparation. This extreme surface sensitivity further complicates analysis of spectra, as the surface electronic structure may be quite dissimilar to the bulk, and may contribute significantly to the data. Both experimental and theoretical aspects of photoemission spectroscopy have been treated in a number of thorough reviews (Cardona and Ley 1978a,b, Hfifner 1996, Kevan 1992, Smith and Himpsel 1983), to which the interested reader is referred for more extensive discourse. Here our discussion will be limited to a brief overview of the aspects of PES methodology relevant to angle-resolved studies of the high-temperature superconductors. A simple intuitive understanding of photoemission begins with the three-step model of Spicer and Berglund (Berglund and Spicer 1964) in which emission is broken up into distinct excitation, transport, and detection steps. A schematic illustration of this process is shown in fig. 4. Excitation involves occupation of an excited crystal state by the photoelectron within the bulk of the material via a crystal momentum (k) conserving direct transition. This electron is then transported through the surface barrier and propagates to the detector. Within the framework of the independent-electron

398

M. SCHABELand Z.-X. SHEN

approximation the excitation process can be viewed as the transfer of an electron from an occupied to an unoccupied single-particle state. In this picture, the energy distribution of electrons obtained in a photoemission experiment corresponds to the density of occupied states modulated by transition matrix elements from the photoabsorption crosssections and superimposed on a smooth background arising from inelastic scattering processes. For materials which are well described by an independent-particle picture, and for photon energies sufficiently high that the photoelectron final states are planewave-like, this interpretation is sufficient to describe the essential features of the experimental spectra. Unfortunately, while heuristically useful, the three-step model lacks the sophistication to describe details quantitatively. In materials such as the cuprates, in which electron correlations play a crucial role in determining the electronic structure, localized interactions between the photoelectron and the photohole can significantly modify spectra, leading to satellite features which do not appear in band-theoretical calculations. Nevertheless, it is often fruitful to interpret photoemission results as far as possible within the framework of band theory, as unconventional behavior beyond that predicted by mean-field calculations provides crucial clues to the poorly understood aspects of a materialN electronic structure.

2.1. Angle-resolved photoelectron spectroscopy An extension of photoemission spectroscopy which has proven particularly powerful as a tool for experimental determination of band structure is angle-resolved photoelectron spectroscopy (ARPES). In an ARPES experiment, the distribution of photoelectrons from a crystalline sample is measured as a function of emission angle relative to the crystal axes (fig. 5). Conservation of crystal momentum, k, parallel to the surface combined with energy conservation establishes a correspondence between the angles (0e, q~e) and the components of kll. Due to the non-conservation of k±, in general it is not possible to extract complete information on the k-dependence of the bulk electronic structure. For this reason, extraction of band dispersions from three-dimensional materials is complicated, although judicious use of the ability to control photon energy can provide relative kz information as well. Fortunately, in highly two-dimensional materials such as the cuprates there is very little c-axis (interlayer) dispersion, making the Brillouin zone nearly independent of kz. In this case the perpendicular momentum component becomes irrelevant and the band structure is completely determined by the parallel components. The technique has been quite successful in measuring band dispersions in many materials including simple metals, semiconductors, transition-metal oxides and chalcogenides (Htifner 1996, Shen and Dessau 1995, Smith and Himpsel 1983). It has been particularly powerful when used in conjunction with ab initio total energy calculations (Callaway and March 1984, Jones and Gunnarsson 1989), and is the only direct experimental means of probing the full band structure of solids.

ANGLE-RESOLVED PHOTOEMISSION STUDIES OF UNTW1NNED YBCO

399

Fig. 5. A schematic illustration of an angle-resolved photoemission experiment. An incident photon, with wavevector p and polarization E, strikes the sample with polar incidence angles (0o,0p) relative to the crystal axes. In practice the light source is generally fixed relative to either the crystal or the detector. However, the ability to vary the photon polarization from synchrotron sources provides a powerful tool for obtaining information on the symmetries of electronic states. By moving the analyzer or the sample (depending on the details of the experimental apparatus), photoelectrons leaving the surface at polar angles (0~,~) are collected by the spectrometer; the component of their crystal momentum, k, parallel to the sample surface is strictly conserved, allowing accurate determination of the two-dimensional band structure. 2.2.

Resonance photoemission

R e s o n a n c e p h o t o e m i s s i o n (RESPES) is a powerful tool for extracting species specific spectral i n f o r m a t i o n in materials with localized orbitals. In the resonance process, two distinct q u a n t u m pathways exist for creation o f a photoelectron: (i) the direct p h o t o e m i s s i o n process in which a valence electron is directly emitted, a n d (ii) the A u g e r c h a n n e l in w h i c h a photoemitted core electron occupies the partially filled valence orbital, followed b y a K o s t e r - K r o n i g A u g e r decay i n which the core hole is occupied and the valence electron photoemitted. Q u a n t u m interference b e t w e e n these two channels n e a r the resonance threshold leads to strong cross-section variations for the localized states, allowing their signal to be differentiated from the contributions o f other atomic species. For the case o f copper, the two processes are: 3p63d 9 + h v ~ 3 p 6 3 d 8 + e ,

(i)

and 3p63d 9 + hv--+3p53dl°---+3p63d 8 + e .

(ii)

Clearly, the latter c h a n n e l is inaccessible until the p h o t o n energy is sufficient to create the core excitation, hv,-~E3p, and will o n l y have an appreciable effect near the absorption

400

M. S C H A B E L and Z.-X. SHEN F-.w.--r-w~-~r---T--~--r*--r--r--~r--~.--r--.r-w-~V-

I I5~v

124eV 124eV-115eV

~o

,~ ~

~---r--~-r-.-c~

a)

T~--l----~r---~-V--

~sev

t 24eV 124eV-115eV

•

•

~

b)

D~

o

o

z

z

Pr 0.22

3/3/95 20.0

1,5.0

10.0

5.0

Binding energy (eV)

Pr 0.50

3/17/94

J---LJ.. !~.I__±.J~_.L~.- J_L_~__~-.a...L_J_.~...J__.~...L..

0.0

_~_-~_.±~-L-J._~a..-L-~..~...L~.~--.~.~a-_~J.~-~-..a..a_J_

20.0

15.0

10.0

5.0

0.0

Binding energy (eV)

Fig. 6. Photoemission EDCs taken on twinned crystals of Yj vPryBa2Cu306.95, for x=0.22 and 0.50, below (hv = 115 eV) and above (h~'= 124 eV) the Pr 4d resonance threshold. The Pr contribution to the density of states is clearly revealed in the near-E~ region of the difference curves, scaling with the doping level; slight antiresonance behavior is seen in the Ba 5p core levels. resonance (Davis 1982). It is this fact which provides chemical specificity; a resonant enhancement o f the emission will occur for photon energies which depend on the element-specific absorption threshold for core hole creation. The ability to separate the contributions o f various species to the overall density o f states in experimental spectra is of especially great value in studies o f materials with many different constituent elements, such as the high-temperature superconductors. An excellent review of the RESPES technique as applied to studies of TMO, rare-earth oxides, and heavy-fermion materials has been given by Allen (1992). Figure 6 shows application of the the RESPES technique to samples of Y123 doped with varying amounts o f Pr, which substitutes for Y in the inter-bilayer site and is known to cause dramatic suppression of Tc (Schabel et al. 1995). The clear presence of significant Pr 4 f weight near EF in this material is evidenced by a peak at low binding energy in the difference spectra which scales with Pr doping. 2,3. Structure of Y123 Stoichiometric YBa2Cu307 6 ( 6 = 0 ) is a bilayer euprate which is slightly above optimal hole doping, with a superconducting transition temperature a few degrees below the maximum value of 94 K found for ~ = 0.05. Its crystal structure (fig. 7) has, in common with all other HTS, planar CuO2 layers, with each bilayer separated by an atom of Y (or other rare-earth element). Unique to Y123, the interlayers are formed from quasi-one-

ANGLE-RESOLVEDPHOTOEMISSIONSTUDIES OF UNTWINNEDYBCO

Ba

Y

401

Fig. 7. The crystal structure of Y123, showing the pyramidal units formed by the planar Cu(2) atoms, each surrounded by four planar 0(2) and 0(3) atoms and one out-of-planeapical 0(4). The large spheres represent the interbilayer Y atoms, and the smaller spheres the Ba atoms which fourfold coordinate with the apical oxygen. The chains running parallel to the b-axis, with each Cu(1) twofold coordinated to chain O(1) and twofold coordinated to apical 0(4), are shown as vertically oriented plaquettes.

dimensional CuO3 chains (stabilized by Ba) which run parallel to the b-axis, inducing significant orthorhombicity, as well as complicating interpretation of physical properties. Under normal growth conditions, crystals o f Y123 have mixed domains with alternating orientation axes for the quasi-1D chains; such crystals are normally referred to as twinned. These twinned samples may be oriented to form untwinned crystals by an annealing procedure in which the crystals are simultaneously heated and subjected to uniaxial stress under a controlled oxygen atmosphere. This process can, in the best samples, produce almost perfectly oriented crystals with a single uniform domain. The close chemical and structural resemblance o f the interlayers and planes also hinders spectroscopic studies. in the notation o f the figure, there are two inequivalent copper atoms: chain Cu(1), and plane Cu(2). O f the four inequivalent oxygen sites, O(1) is referred to as the chain oxygen, 0(2) and 0(3) are plane oxygens, and 0(4) is the apical oxygen. Structural studies o f Y123 also reveal a clear buckling, or dimpling, of the CuO2 bilayers, with the 0(2) and 0 ( 3 ) atoms displaced out of the plane by Cu(2) toward the intercalated Y atom (Andersen et al. 1991, 1994, Pickett 1989), as well as static or extremely soft bending modes for the chain oxygens (Pickett 1995, Schleger et al. 1994, 1995, Sharma et al. 1996).

402

M. SCHABEL and Z.-X. SHEN

As mentioned above, Y123 in stoichiometric form is nearly optimally doped, with maximal Tc attained by slightly reducing the crystals to decrease their oxygen content. It is well known that the reduction process leads exclusively to loss of the chain O(1) atoms, leaving the structure of the CuO2 planes unaltered while simultaneously introducing defects to the CuO3 chains. For 6 ~ 0.4 there is structural evidence for an ordering transition to the Ortho-II phase, in which O(1) vacancies preferentially order in alternating chains, leaving neighboring chains intact. This transition, closely coinciding with the observed orthorhombic to tetragonal structural phase transition found near 06.6, may also be responsible for the plateau found near this doping level in the Tc vs. 6 curves. Further deoxygenation introduces defects in all chains, and the material ultimately becomes a tetragonal charge-transfer insulator, YBa2Cu306, in which C u Q dumbbells replace the chains. Y123 can also be doped with a wide variety of impurities, including Zn and Ni which substitute for Cu(2) and suppress Tc dramatically, Co which replaces Cu(1), and Ca or various rare earths for Y. The possibility to precisely manipulate the chemical composition and impurities in this material is unique among the cuprates, and is one of the principal reasons Y123 is an attractive system for scientific investigations. Substitution of the inter-bilayer Y in Y123 with various other rare-earth elements is found, in most cases, to have essentially no noticeable effect on the superconducting transition or properties. Crystals of the form RBa2Cu307_ 6 (R123) are isostructural with Y 123, having virtually identical lattice parameters, superconducting transition temperature, and physical properties. Among the fourteen rare earth elements, there are two notable exceptions to this rule: Ce and Pr. a Of these, the former does not crystallize in the Y123 structure, phase-separating into simpler constituent compounds instead. Surprisingly, Pr123 is isostructural but manifests completely different physical properties. Optimally doped Y123 is metallic, with a transition to the superconducting state at 94K, and manifests antiferromagnetic alignment of Y moments at extremely low temperatures (-1 K), while the Pr material is an insulator which exhibits no tendency to superconduct, with Pr moments aligning at 17K. Experimental results have been reviewed in a thorough article by Radousky (1992). Doping of Y123 with various amounts of Pr to form Yl-xPrxBazCu307 6 (YPrI23) results in a rapid suppression of Tc with x, reaching OK at x=0.55. The absence of any significant tailing in magnetization measurements and the narrowness of the superconducting transition in resistivity measurements demonstrates that the Pr atoms are homogeneously distributed rather than phase segregated into regions of high and low Pr concentration. This observation supports a continuous microscopic mechanism of superconductivity suppression, which is of obvious importance as a clue to understanding the pairing mechanism in the cuprates, as it must depend on specific details of the electronic structure and chemistry of Pr. Early arguments centered on the simple possibility that Pr, substituting for y3+, takes on the Pr 4+ oxidation state, thereby forcing t The Tb compoundis difficult to synthesize in crystalline form, but thin films have been grown successfully. Pm, which is unstable and radioactive,has not been studied.

ANGLE-RESOLVEDPHOTOEMISSIONSTUDIESOF UNTWINNEDYBCO

403

the material into a highly underdoped insulating state (Liang et al. 1987, Okai et al. 1988, Soderholm et al. 1987). Numerous spectroscopic techniques, particularly X-ray absorption spectroscopy (XAS), electron energy loss spectroscopy (EELS), and PES, which are capable of directly comparing measured spectra with those from reference samples of known oxidation state, have ruled out this mechanism, clearly demonstrating that the Pr remains in the 3+ state (Soderholm and Goodman 1989). Other theoretical arguments include magnetic pair breaking (Guo and Temmerman 1990), changes in the relative chemical potential of the holes in the CuO3 chains and CuO2 planes (Khomskii 1993), and hybridization of the low-lying Pr 4f states with the O 2p orbitals on the cuprate planes, thereby localizing the normally mobile ligand holes (Fehrenbacher and Rice 1993, Liechtenstein and Mazin 1995). To date, none of these proposals have compellingly explained the behavior in this material, and the problem appears to remain an open one. Recent experimental work has also demonstrated superconductivity in these materials under growth conditions in which the c-axis lattice constant is reduced from its bulk value. It seems certain that a detailed understanding of this phenomenon wild provide significant insight into the mechanism of superconductivity in cuprates. 2.4. SurJhce termination of 11123 The low-temperature STM studies of Edwards et al. (1992, 1994) clearly resolve regions of chains interspersed with indistinct stepped regions whose height is consistent with that expected for BaO planes. These results provide strong evidence for a mixed CuO3/BaO surface termination in cleaved Y123 crystals. Photoemission corroborates this, revealing a strong Ba 5p surface-shifted component, shown in fig. 8, which is indicative of inequivalent bulk and surface Ba coordinations as would arise from a near-surface BaO layer (Halbritter et al. 1988). Resolving a similar shift in the chain Cu(1) levels is unfortunately complicated by the presence of two structurally inequivalent Cu atoms in the bulk unit cell, in addition to two possible inequivalent surface sites and the presence of satellite structures in the Cu core levels. It is well known from studies of the electron attenuation length in photoemission experiments that the mean free path of unscattered photoelectrons at the photon energies used in ARPES experiments is extremely short, on the order of 5 A (Seah and Dench 1979). This fact, and the relatively large c-axis lattice constant of Y123 (11.7 A), combine to make the surface contribution in our spectra a substantial fraction of the total signal. A simple layer model allows us to quantitatively estimate the contribution of the surface and subsurface Ba atoms to the total photoemission intensity (Schabel et al. 1998b). We assume a mixed termination of BaO and CuO3 with equal areas for both terminating surfaces, shown schematically in the inset of fig. 8, make the approximation that the various atomic planes are equally spaced, and regard all subsurface Ba atoms as equivalent to those in the bulk material. In this case, the Ba surface component comes entirely from the BaO surface, giving a surface to bulk ratio for the Ba core levels of 82%, in good qualitative agreement with the nearly equal contributions of the surface (high binding energy) and subsurface (low binding energy) features seen in fig. 8, as well as with the

M. SCHABELand Z.-X. SHEN

4.04 I

,

,

,

,

I

,

L

,

BaO CuO BaO GUy02

,

I

,

,

1

......... ............. -

0~oj ................................ CuO BaO C%02

......................

Cu02

• ,~,-4 M3

©

BaSp't /

°,~ r/b

2©

/

!

-

L ,

i

i ."

J

~

15

,

)

10

i

T

i

~

I

i

,

5

Binding Energy (eV)

,

i

0

Fig. 8. The full valence band of YBCO, showing distinct bulk and surface shifted Ba 5p core levels. While there can be substantial sample to sample variation in the precise magnitude of the surface core-level shift, it can be dearly seen in nearly all spectra, indicating the presence of significant regions of BaO surface termination. The inset shows a schematic of the surface termination of cleaved Y123, with stepped terraces of BaO and CuO~ chains, based on ARPES data and low-temperature STM results.

results of studies of the Ba 4d core levels reported by Veal and Gu (1994). A similar analysis may be performed for the Cu valence state intensities, allowing estimation o f the contribution of "bulk" (subsurface) chain states relative to the surface chain states and bulk plane states. On the general grounds discussed above, we expect the surface contribution to the chain states to dominate the bulk chain emission. Assuming that the intrinsic emission from all states is equal (i.e., ignoring matrix element effects), and modifying the model used above for the Cu states leads to contributions o f 59%, 32%, and 9% for the bulk planes, surface chains, and bulk chains, respectively. Even allowing for an escape depth twice as large (i.e. 10 A) only changes the relative contributions o f plane, surface, and chain to 64%, 18%, and 18%. Based on these estimates, it is likely that the bulk chain features o f the electronic structure only make a small contribution to the measured photoemission spectra.

3. P h o t o e m i s s i o n

studies of the cuprates

Progress in growth o f high-quality single crystals of various high-Tc superconductors, particularly Y123 and Bi2212, was a development crucial to the success o f ARPES

ANGLE-RESOLVEDPHOTOEMISSIONSTUDIES OF UNTWINNEDYBCO

405

YPr123 valence band (hv = 28eV) - - r - - r - v - w - - V - X - ~ - r - - r ' - v - - c - r - w ~ w"v~.[1

L

.=, .,-~

1

4..a

0

b9

Ba 5:

a

?

°t

?

D

Z

%

5.0

10.0

15.0

20.0

Kinetic energy (eV)

25.0

Fig. 9. Valence band spectra fi'om two twinned Y0s0Pr~s0Ba2Cu3069s single crystals. The bad cleave has the characteristic contaminant peak at 9.5 eV binding energy filling in the region between the valence band and the Ba 5p core levels, while the good cleave shows little weight in that energy region, as well as marked enhancement of the valence-band intensity relative to the Ba cores.

studies of these materials (Lin et al. 1992, Mitzi et al. 1990). Early work on lowquality polycrystalline samples, which suffered from indeterminate surface structure, contamination, and high defect and impurity densities, routinely failed to manifest any Fermi edge structure at all, even when transport measurements indicated clear metallic character. The principal hallmark o f low-quality or contaminated sample surfaces for the cuprates is the appearance o f a large, indistinct feature near 9.5 eV binding energy in the photoemission data, an example o f which is shown in fig. 9, while the presence of weight in a coherent dispersive peak near El~ is indicative o f a good cleave. Another critical development was dramatic improvement in the spectrometer energy resolution, from typical values of 200 meV commonly considered adequate in semiconductor surface physics to values of 2 0 m e V or less achievable today. Because superconductivity is a phenomenon whose energy scale is defined by some small constant times k~3Tc, resolution comparable to or smaller than this characteristic value is essential for direct observation o f modifications in the near-EF spectral weight associated with the onset o f superconductivity. O f the various cuprate materials, Bi2212 is by far the most extensively studied, owing to the availability of large, high-quality single crystal samples, the ease with which it

406

M. SCHABELand Z.-X. SHEN

cleaves, and the existence of a reproducible unreconstructed surface uniformly terminated by comparatively inert BiO planes. Also, as the first high-temperature superconductor in which a superconducting gap was consistently observed by photoemission, an immense amount of work has been devoted to this aspect of the electronic structure. Details relevant to photoemission studies of optimally doped Bi2212 in both the normal and superconducting states have been discussed extensively in reviews by Shen and Dessau (! 995) and Loeser et al. (1994). Significant efforts to investigate the electronic structure of Y123 have also been made, although the lack of a natural charge-neutral cleavage plane poses difficulties. Unlike Bi2212, all possible cleavage planes result in a mixed termination of two inequivalent surfaces. In addition, it appears that oxygen loss or disorder leads to formation of an insulating surface, even for optimally doped crystals, unless samples are cleaved and maintained at temperatures below 50K (List et al. ~,,o~. Unfortunately, this precludes cycling of temperature through the superconducting transition, making identification of superconductivity-related modifications of the spectral weight even more difficult. Much of the experimental methodology and relevant results from photoemission studies of twinned Y123 have been discussed in a thorough review by Veal and Gu (1994). Here we focus primarily on untwinned crystals and the low binding energy region within 1 eV of EF; details of the valence-band electronic structure, including the peak appearing at 1 eV binding energy, do not appear to have the same strong dependence on the a/b-axis asymmetry (Tobin et al. 1992). In the untwinned samples, orientation of the fourfold axis of samples is performed ex situ by Laue diffraction. Due to growth asymmetry, the chain axis may be determined visually as the long axis of the crystal; the accuracy of visual orientation is verified with X-ray diffraction. Typical studies are performed on crystals of extremely high quality (typical transition widths of 0.25 K), cleaved and maintained in UHV with a chamber base pressure less than 5× 10 -11 torr at a constant temperature of 20 K to avoid surface degradation (Edwards et al. 1992, List et al. 1988). Spectra are normally acquired at an energy resolution of 50 meV or better, with an angular acceptance of ± 1o 3.1. Normal-state electronic structure

At the coarsest level, the poor state of overall understanding which applies to transitionmetal oxides in general also plagues the high-temperature superconductors. In this sense, the crucial problem which faces scientists studying the cuprates is the normal-state electronic structure of these materials and how it relates to and deviates from conventional Fermi liquid theory. As a collective property which appears at low energy scales for relatively well-understood reasons, pairing superconductivity should emerge naturally as a low-temperature collective phenomenon once the ground-state properties have been clarified. Conversely, it may not be comprehensible without a thorough understanding of the anomalous normal state from which it arises. In addition to studies of the HTS in the temperature regime above their superconducting transition, there have been recent, fruitful efforts to study prototypical model compounds such as the insulating cuprate

ANGLE-RESOLVED PHOTOEMISSION STUDIES OF UNTWINNED YBCO '

-

I

'

I

"

I

'

[

'

I

,

I

Bi2CaSr2Cu208

-

hv

,

/'

'

407

I

=

I

,

I

,

I

,

I

,

t

i

I

i

I

-12 -8 -4 0 Energy Relative to E F (eV)

Fig. 10. Valence band spectra from Bi2212 single crystals above and below the Cu 2p resonance. An anomalous enhancement of spectral weight lying between 9 and 13 eV binding energy is a satellite feature due to manybody effects.

S r 2 C n O 2 C 1 2 (Wells et al. 1995) and the quasi-lD cuprate Sr2CuO3 (Kim et al. 1996). These studies may help to elucidate the fundamental properties of the insulating ground state out of which cuprate superconductors are formed, and give further insight into the relevant low-energy degrees of freedom for theoretical models. Much current interest, both experimental and theoretical, is centered on the difficult topic of the transition from insulating parent compound to superconducting metal, as well as on the effects of reduced dimensionality in strongly correlated chain and ladder compounds (Dagotto and Rice 1996, Emery and Kivelson 1993, 1995, Rice et al. 1994a,b). One indicator of strong electron correlations in solid-state systems is the presence of high-binding-energy satellite lines corresponding to localized excitations. The observation of such features in photoemission spectra of the cuprates provides strong evidence supporting the contention that Coulomb interactions play a critical role in determining the electronic structure of the CuO2 planes which give rise to superconductivity. Satellites cannot be explained by conventional band theory because they arise from local excitations rather than itinerant electron states, but are evident in cluster calculations which explicitly include correlation effects (Tjeng et al. 1991, Zaanen et al. 1985). Photoemission experiments by Shen et al. (1989) find clear evidence of a high-energy satellite feature associated with the Cu 3d orbitals in Bi2212, with the spectral weight lying between 9 and 13 eV binding energy, as shown in fig. 10, anteceding similar measurements by Fujimori et al. and others on Y123 and LazCuO4 (Arko et al. 1989, Fujimori et al. 1987, List et al. 1989). Perhaps the most interesting characteristic of the metallic cuprates is the coexistence of these Cu satellites with low-energy quasiparticle states. This behavior correlates with their transitional nature, lying between conventional Fermi liquids and the

408

M. SCHABEL and Z.-X. SHEN

charge-transfer insulators from which they are derived, and lies at the heart of current debate over the appropriate theoretical framework for their description. Current debate over the nature of the quasiparticle states in the cuprates and the pairing mechanism in the superconducting state has emphasized the uncertain applicability of conventional Fermi-liquid theory and focussed attention on unresolved questions in ARPES studies of materials considered to manifest normal Fermi-liquid properties. In particular, work by Claessen et al. (1992, 1996, 1997) has demonstrated that, in the layered chalcogenide TiTe2, the ARPES lineshapes may be consistently and quantitatively fit with a model based on a Fermi-liquid-like self-energy. In marked contrast, very little progress has been made in the fitting of lineshapes in the high-To materials despite repeated attempts (Campuzano et al. 1991, Liu et al. 1991, Olson et al. 1990a). The universally observed presence of a significant fraction of the total spectral weight at low energies in apparently incoherent structure continues to be an enigmatic clue to the nature of their low-energy excitations. The vast majority of recent photoemission experiments on the cuprates have focussed on the narrow, dispersive features found near the Fermi level, first observed by Arko et al. (1989). These states arise in a small foot in the lowest 1 eV below EF, and grow more prominent as the doping level approaches the optimal value. Careful measurement of the amplitude and peak position of these quasiparticle features allows reconstruction of the band structure and the two-dimensional Fermi surface topology, and can reveal the presence and magnitude of a superconducting gap. Transport, optical, and electrical properties in the cuprate superconductors all manifest anomalous behavior; these also must be closely linked to the spectrum of low-energy excitations, since the relevant states lie within a few kB T of EF. In Y123 and Y124 an extended van Hove-like singularity (EVHS) is associated with an extremely narrow, resolution-limited peak (A ~< 10 meV) along the FX/FY direction in twinned samples. This band appears approximately 250 meV below Ev at F, dispersing to within 10 meV (Y123) or 19 meV (Y124) of the Fermi level at approximately the midpoint between F and X/Y where it remains out to the Brillouin zone boundary. Along the zone boundary from X/Y to S a band is observed to disperse rapidly upward away from EF which combines with the flat band to form the EVHS saddle point. A second singularity is also observed, but it lies significantly deeper in the valence band, with a binding energy of 115 meV, where it should not have any dramatic effects on the DOS. It should also be noted that the EVHS persists even in oxygen-deficient samples (6 = 0.5) with substantially depressed Tc's and significantly different physical properties (Liu et al. 1992a,b). The near-EF singularity is one of the most intriguing features of the electronic structure of Y123. This peak appears to be essentially non-dispersive parallel to the FX/FY direction and strongly dispersive along XS/YS, forming the EVHS. It also manifests unusually strong photon energy dependence, being seen most strongly at h v = 17 and 28 eV, and also at 50 and 74 eV (Gu et al. 1993). In their early study of an untwinned single crystal of Y123, Tobin et al. (1992) observed this feature only near the Y point, and concluded that it was correlated with the CuO3 chains. A similar conclusion was drawn for Yl24 crystals by Campuzano et al. (1992), based on the relationship between

ANGLE-RESOLVEDPHOTOEMISSIONSTUDIESOF UNTWINNEDYBCO

409

X/Y asymmetry and the uniformity of the cleaved sample surface, and for Y123 based on the results of oxygen doping dependence (Gu et al. 1993). However, observation of the peak along FY placed the non-dispersive direction parallel to the chain axis, rather than perpendicular as would be expected for quasi-lD states. This counterintuitive result is also inconsistent with both local density approximation (LDA) predictions of a chainderived Fermi sheet parallel to the FX axis and the chain Fermi surface determined from positron-annihilation studies of untwinned crystals (Shukla et al. 1995). Subsequent photoemission work interpreted the EVHS as a CuO2-plane-derived feature arising from the chain-induced orthorhombic distortion, and implicated the resulting singularity in the density of states as a possible origin for the high transition temperature (Abrikosov et al. 1993, Gofron et al. 1993, 1994). Much of this work was driven by the observation of an extended flat band near the Fermi energy in ARPES studies of Bi2212, even though that feature derives from a much broader band in the normal state and exhibits clear and dramatic modification in lineshape on passing into the superconducting state. More recent results, obtained on untwinned Y123 crystals of exceptional quality, lead to a different picture of the electronic structure in Y123 and Y124, more consistent with earlier work on twinned crystals, and reveal numerous previously unobserved features (Schabel et al. 1997, 1998b). These data indicate that the near-EF electronic structure along the b-axis is strongly masked by the presence of a narrow and intense surface resonance, or surface chain feature (SCF), arising primarily from the quasi-lD CuO3 chains, making untwinned single crystals essential to have any hope of resolving a superconducting gap. In addition, varying the sample orientation with respect to the photon polarization is crucial for a complete understanding of the electronic structure of Y123. 3.2. Surface chain .feature The presence of a feature at the Y point, absent at X, does not of itself imply chain parentage (due to the orthorhombic distortion of the underlying lattice), but the extremity of the asymmetry and apparently one-dimensional behavior are difficult to reconcile with the 2D planar structure of the CuO2 sheets. Also difficult to understand, assuming that the orthorhombic distortion is responsible for the asymmetry, is the nondispersive nature of the SCF with photon energy. A number of additional arguments, delineated below, provide strong support for the contention that the SCF is indeed derived from the CuO3 chains. Figure 11 shows ARPES EDCs taken at the X and Y points, with hv=28eV, for the three distinct polarization geometries; panel (a) corresponding to E It b, panel (b) to E i b , and panel (c) to E IIFS (02, 04, and OI, respectively). It is important to recognize that, while the absolute value o f the intensity axis is not significant, the relative amplitudes between spectra are, as they are scaled to the integrated valence-band emission, and may be compared quantitatively. To the extent that the chains are truly onedimensional, photoemission spectra taken with photon polarization parallel to the b-axis should maximize their contribution to the signal, those with perpendicular polarization should be essentially free of chain signal, and those with intermediate polarization should

410

M. SCHABEL and Z.-X. SHEN

o.7 o.4~

,,:,

~,

0.6

o.,

0.8

~'

_~ g

~',,,:

,

x

i

:

'i ,~

0.35

g

.~ 0.6

0.3

E

(3-

/

'7'v"

~'i

,

-

z

0.2

o

-0.8

~

~.,~,o

,x

:

-0.4

,

-0.2

Binding Energy(eV)

,

0

o.1 I

x

oo51. ~ -0.8

I

\

'\

Y

-0.6

0.3 \

,~

" ....

v

\

0.4

0.25 ~1¢\,,

Y

¥ ,\

x

,~

-0.6

P

-0.4

02

,

-0.2

Binding Energy(eV)

,~'°

0

0.1

o

-0.8

, i

"'.,--.:#i s

s

~

-016 -0'.4

-0'.2

/~,%~

0

Binding Energy(eV)

Fig. l 1. Photoemission EDCs taken near the Fermi energy at the X and Y points of a single-crystal sample, for photon polarization (a) E II b, (b) E • b, and (c) E II FS.

mix the two. Contrasting with this, the predominantly dx2 y2 Cu 3d states hybridized with O 2p states to form the plane bands constituting the near-EF structure should be comparatively unaffected by the 90 ° rotation, other than experiencing slight asymmetry due to the orthorhombic structure and the normal selection-rule effects for electron emission. Examination of the spectra at Y dramatically demonstrates the dependence of the SCF on sample orientation; the spectrum with E II FY shows enormous enhancement, while the perpendicular geometry reveals nearly complete suppression, and the 45 ° spectrum shows almost exactly half the intensity of that in 02, as expected from geometrical arguments. Furthermore, in addition to its factor-of-eight diminution in intensity, the small residual peak seen at Y in 04 is qualitatively unlike the SCF, being much broader and lying at measurably higher binding energy, and presumably originates in the planes. This quasiparticle behaves in a manner qualitatively similar to that seen in Bi2212. A third feature of the electronic structure, which appears weakly as an extremely broad peak at a binding energy of 650 meV, is essentially unaffected by the polarization component parallel to the chains so it must also derive from the underlying CuO2 plane states. Also important is the comparatively weak dependence of the overall intensity away from EF on geometry, all three EDCs at Y lying between 0.25 and 0.45 at the high binding energy limit. In fig. 12 we plot the difference between normalized spectra at the Y point in the geometry in which polarization is parallel to the chains and that in which it is

ANGLE-RESOLVED PHOTOEMISSION STUDIES OF UNTWINNED YBCO

411

.=~ 7_=

Y point, 28 eV 02 - 04 i

i

i

0.6

0.4

0.2

0.0

Binding energy (eV)

Fig. 12. Normalized ditt~rence between spectra taken near the Fermi energy at the Y point for photon polarization parallel to and perpendicular to the CuO 3 chains.

perpendicular to them. What is observed strikingly confirms the arguments above; the only significant difference between the two in the 0.5 eV below the Fermi energy is in the narrow peak lying at a binding energy of 13 meV, as reported in previous studies of the EVHS in Y123, and with a width which is almost entirely comprised of the instrument contribution. It is important to note that the two spectra were taken on the same sample, the only difference between them being the photon polarization relative to the chains. The extreme narrowness of this feature is in stark contrast with the other dispersive bands seen on the same sample, all of which are broader by a factor of 10 to 20. By this coexistence of both very narrow and very broad features in the photoemission spectra from one sample we are able to infer that the broad peaks are intrinsic in origin, and do not stem from poor sample quality or energy resolution. Turning to the spectra at X, we immediately note the absence of any features in the near-EF weight of comparable intensity to the SCF in any polarization orientation. As for the Y point, the high-binding-energy intensities are only weakly dependent on geometry, although the matrix elements at X appear to lead to overall enhancement for emission from this region of k-space. Also as seen at Y, a broad higher-bindingenergy feature (around 400 meV) with little dependence on the chain component of E is found. It is important to also consider the (likely) scenario of residual twinning of the crystals at the few percent level, as the detwinning procedure is never entirely perfect. In the case of a lightly twinned crystal, the intense feature near Y will be

412

M. SCHABELand Z.-X. SHEN

mirrored at X, albeit with its intensity significantly reduced. Nevertheless, because this structure is so intense in the parallel polarization geometry, even a few percent can cause a significant and spurious addition of spectral weight in the energy region immediately below EF. The absence in Bi2212 of a feature comparable in width and anisotropy to the SCF provides an additional clue; the extended flat band in the Bi compound exhibits substantially greater dispersion, is observed symmetrically along both of the cubic axes, and is much broader than the SCF in Y123 (Dessau et al. 1993). Conversely, a similar SCF feature, with even greater intensity, is found in Y123's sister compound, Y124 (Gofron et al. 1993); the peculiarity of the SCF to compounds possessing CuO3 chains is itself highly suggestive. Comparisons of the relative peak to background ratio, Rp (defined as the ratio of the peak height at Y to the incoherent spectral weight 0.2 eV below EF), from various twinned and untwinned samples of Y123 and Y124 show that Rp in untwinned Y123 is roughly twice that oftwilmed samples, with untwinned Y124 twice again as large. This variation is easily explained by two facts. First, in a twinned sample the intensity from the Y point will be evenly distributed between X and Y. Second, as previously discussed, the principal structural distinction of the Y 124 material is that it contains twice as many chains per unit cell and is natm'ally (partially) untwinned (Campuzano et al. 1992); it is expected that the cleaved surface should have a similarly disproportionate number of exposed chains, leading to an enhancement factor of two in samples with uniform cleavage planes. The specific ratio will, of course, depend on the sample and is expected to exhibit cleave-to-cleave variation due to differences in the BaO/CuO3 distribution at the interface. Finally, Pr doping, which substitutes Pr for the interplane Y atom, has the unique property among the substitutional rare earths of strongly suppressing the superconductivity of Y1 xPrx123, inducing a metal-insulator transition at a Pr concentration of x ~ 0.55. Because the Pr atom lies between the CuO2 planes and is quite isolated from the chains, it would be expected to have a comparatively large effect on plane-derived states and to only weakly affect chain features. In studies of twinned crystals of YPr123 (Schabel et al. 1995) it has been observed that the most dramatic changes in the electronic structure at low doping are indeed observed along the FS line, the main feature of which derives predominantly from plane states and is unaffected by the twinning. In contrast, the SCF is clearly present, even for x = 0.22, where Tc has been suppressed to 60 K. Our results strongly mimic changes observed in the near-EF features of oxygen-deficient samples, indicating that Pr alters the relative charge balance between chains and planes, but does not affect the presence of the SCF at moderate doping levels. Taken together, this evidence provides a consistent and compelling case for association of the SCF with a surface chain state which is uncharacteristic of the bulk electronic structure of Y 123, and which, by virtue of its intensity and proximity to the Fermi energy, strongly masks the presence of any underlying states or gaps in the k-space region near the Y point. Furthermore, except under the most favorable conditions and with almost perfectly detwinned samples, highly intense echoes of the SCF will appear near EF at the X point as well, contaminating data from this region as well.

ANGLE-RESOLVED

PHOTOEMISSION

STUDIES

OF

UNTWINNED

YBCO

413

S quasiparticle dispersions

_4

a

I<

b

1 ....

-N~"

•

~

1.00

0.39 0.36 0,32 0.29 0.25 0.21 0.18

..... ~'~

,%.

,

~0.14 0.07

oe o4 o2

01k

o v.vu

06 o4 o2

E

Ofk E .

• •

,

•

•

•

" • •

• .

01k

0.50 0.47 .44 0.41 0.38 0.34 0.32 0.30 ~ 0.28 0.25 .22 .19

.

.

03k

0.53 0.47 0.44 0.41 0.38 0.34 0

oe 04 02

0 ~.~u

E

~o~ -04 o~

03k

d ~ 0.00

E

02b

~

0.80 0.60 0.50 0.45 0.40 • 0.35 • • 0,30 • 0.25 O.20 0,15 0.10 0.00 .o~ o~ o3 o

E

04a

0.90 0.70 0.50 0.45 0.40

hv = 28 eV

0.35 0,30 0.25 0.20 0.15 0.10 0.00

E

.

.

3

2 0.26 0.25 0.22 0.19

.16

0.16

.13 .09

0.13 0.09

.0060

0.06 0.00

•

E

0 u.vv

1.00

•

-o.6-0.4-o.2

01k E

o

"% ~',w,,',v~

),83 ).67

~

' \

1.00

~ "

).50 ).42 ).38 ).33

5 ~

\ ' / ~ 0"29 • 0.25

\ ~

~ "J0.17

hv = 21 eV

10.13 ~"

" 10.08

o~ .o'4 -o~

04a

~ '0.00 E

Fig. 13. Energy distribution curves along the FS symmetry line. Identifiable quasiparticle peaks are indicated by circles, with the k value of each EDC given as a fraction of the total distance between high symmetry points.

3.3. B a n d structure Band dispersions along high-symmetry directions in the Brillouin zone have been presented in a number of studies of Yl23, but these have been tainted by the mixing o f inequivalent octants of the zone from twinning. Recent work on untwinned crystals has established a complete set of data for the high-symmetry directions FS, FX, FY, XS, and YS, in a number o f polarization orientations with a photon energy of 28 eV, along with somewhat less complete data at 21 eV (Schabel et al. 1997, 1998b). Representative dispersion curves for FS, FX and FY are shown in figs. 13-15. In each set of energy distribution curves identifiable quasiparticle peaks are marked with a circle. Despite the intrinsic breadth of these features (comparable to the dispersive plane bands seen in photoemission data on Bi2212), statistics are sufficiently good to allow accurate identification of the centroid in most cases.

414

M. SCHABEL and Z.-X. SHEN

X quasiparticle dispersions

1.00

1.00 1.00 ~

).93

),71

!,O0 ).86

\ . \ \ '-4

),71

~.57

\'I, X_~ I

),43

h

\

\,~-t

: \',,\\"-4. ),43 I \ \ \ \

~

\ ).29

).29

~ ~"

).21

;.14 .o; .o;: .o?

Ofk

o

"'

.......

o~ o,~ o~

01k E

03 k

o

).00

~''"

6~ 0.00

03k

E

).80

,,,~,~,~

0.83

).70

, "~,~..~,~

0.75

),60

0.67

).50 ).45

0.58

).40

~~

~"

).35

, ~

,.

02b

0.42 0.33 0.25

1,20 o.~-o"-o~

0.50 ,

).30 ).25

0.20 0.15 0.10

-o~ o~ o~

E

0.92

0.80 0.70 0,60 0.50 ~ & "C',-...Y • 0.45 "v,~. \ \g. 0,40 . '~..- \ \ k . ~ 0.35 . . . . 0.80 ~ a'd' \ ~" 0.25 \.k ",4

).14

).00

E

.....

•

o

~

"

0.17

"

1.10

0.08

).O0

.o~ -o.'4 .o.'~ o 0.00

E

045

£

Fig. 14. Energy distribution curves along the [ ' X symmetry line. Y quasiparticle dispersions . . . .

-..,._ . , z 4 1.00 , "\ 0.86 , . ~ ~ . , , j t ~ " 4 0 . 8 8 ", ~

0.79 ~ ~

' - ~ / ...t. ~ - ~ A

~ Y

T I

01 k

3.93

0.64

058

0.57

0.50

3.57

\\~

D.44

,, •

3.38

3.50 " ~ - - ~ ~ A 3.43

0.50

~'~¢,

0.43

tn ~

0.36

I1[ " ~

\

\

3.71

~\~'~

3.80

3.64

3.32

3.70 ",~ " . , , . . . ~ 2 ~ ~

3.25

0.21

0.19

0.14

3.13

3.21 , . . " ~ . `

3.06

3.14

-

-

o~ 0 4 .o'2

01k £

-

o~ 3.00 ~ 4

03 k

o~

o

E

3.00

,~-~/

3.50 3.48

•

3,40 3.35

"-,--~-,-~'x

3.36

0.29

3.29

"~'~

3.88 : : k ' ~ J 4 " 3.90 385 _ ~

0.63 "x~ "-,~ . . . . . ~ /

0.00

E

~0,75 ~ 1 1

1.oo "~ "-

1.00

0.71

0.07

~'~-:~- ~

"/I L

"

~

~'~

o~ o~-o.~

03k

o

E

3,30 3,25 3.20 3.15 3,10 3.00

1.00 0.90

\ "-*~,~q

0.80

%\~,..~v.

0,70

k

" ~ 0.67 0.58

0.30

0.50 A\, \ * k ~ 0.42 . . . . . 0.33 0.25

0.25 0.20

-o~-o '~ -o~

02b

1.00

~ 0.83

0.50

0.40 0.35 \* \~

~

0.60 0.45

\'~

f

"

"

"

0.17

0.10

0.08

d ~ 0.00

-o.~ .o'4 -o.'2 o~ 0.00

E

04a

E

Fig. 15. Energy distribution curves along the FY symmetry line. An overall picture o f the quasiparticle dispersions along FS is shown in fig. 16, with the energies of features identified in the various panels o f fig. 13 plotted against normalized fraction o f the FS distance. There are several key observations to be made from fig. 16. First, the two clearly resolved quasiparticle features o f plane origin are most sensibly attributed to the bonding and antibonding plane states o f the CuO2 bilayers. The lower-binding-energy feature (the antibonding d* state) dispersing from 0.22 eV at F to a crossing at near 25% o f FS shows essentially no dispersion in kz, with the data points from h v = 21 eV and 28 eV lying on the same curve. In contrast, the higher-energy feature (bonding a state) found at 0.53 eV at F for h v = 2 8 e V and crossing the Fermi surface around 35% o f F S, reveals significantly different dispersion at h v = 21 eV, with the maximum binding energy at F decreasing to approximately 0.42 eV, but with a nearly

A N G L E - R E S O L V E D P H O T O E M I S S I O N S T U D I E S OF U N T W I N N E D Y B C O

415

S dispersion 0 r-r-r'-r-r-'r'~-'r-r-rm-'r~r-'~T~--~"r--r-'r~-'rm

0.1

&

"k

e

Ol, by=28 O[, hv=28 02, by=28

0.~

(hv = 21eV) 0.3

(hv = 28eV)

,~

03, hv=28

A

03, hv=28

v

04, hv=28

v

04, by=28

o

OI, hv=2g

m

02, by=28

o

Ol~ hv=2~

o

O L hv-.21

v

0.'1 by=21 0 4 by=2!

0.6

0

10

20

30

40

50

S/2 Fig. 16. Energy dispersion of quasiparticles along the FS symmetry line. Solid symbols indicate data points taken for hv = 28 eV, while open symbols correspond to h v = 21 e V Band minima are indicated by crosses along the vertical axis, while the corresponding Fermi surface crossings are marked with crosses on the horizontal axis. The thick solid lines are smooth interpolations through the data as guides to tbe eye, while the thin horizontal line indicates the instrument resolution.

identical Fermi surface crossing position. By taking data along identical lines in (kx, ky) at different photon energies we sample different values of k~. Thus, the observed energy dispersion of the plane bonding band with hv is most simply interpreted as a hallmark of c-axis (kz) dispersion of this state. Also significant is the degeneracy of the Fermi surface crossings for the two photon energies, implying that the c-axis dispersion which is large at high binding energies becomes vanishingly small at the Fermi energy. This result is consistent with LDA calculations which predict bilayer splitting of the bonding and antibonding plane states in Y123, although our observed magnitude of 0.2 to 0.3 eV (depending on kz) is approximately one half the value expected from the ab initio results. The E vs. k dispersion relations measured along the FX high-symmetry line are plotted in fig. 17. At high binding energy we see a weakly dispersive feature rising from 0.53 eV at F to 0.43 eV at X. While, based solely on its dispersion, it could be argued that this feature has chain-like behavior, it is clear from its polarization dependence that this

416

M. S C H A B E L and Z.-X. SHEN

X dispersion chain

~

. chain/SS

0.1 l ..

~ e

"

m m w at /a

i l i

~' •

0.2

0.5

.

o.sdg ~. ; ~

0

20

w w o o v¢ v

40

60

80

01. tw=28 0I, hv=28 01, hv=28 02, hv=28 02, hv=28 02, hv=28 03. by=28 03, hv=28 OzL by=28 O4, by=28 04, hv=28 O1, hv=28 OL by=28 ()1, by=28 02, hv=28 O2, by=28 0 [, by=21 Ot, by:=2I 04: hv=2~ ()eL by=2 t

100

X Fig. 17. Energy dispersion o f quasiparticles along the FX symmetry line.

is not plausible, making the most sensible attribution that of the bonding o state of the cuprate plane bilayer. These bonding (o) plane bands are expected to be essentially nondispersive parallel to both FX and FY based on LDA calculations of the band structure, further supporting attribution of this structure to the planes. The quasiparticle originating at lower binding energy at F disperses upward toward the Fermi energy, showing a tendency to flatten out just below EF in the 28 eV data, but exhibiting a clear crossing in the vicinity of the 60% point for h v = 2 1 eV. From the discussion above, this behavior is consistent with a hybrid band arising from mixing of the antibonding plane and chain states. In the k-space region near X, the weakly dispersing shoulder, at an energy of 0.13 eV, forms a van Hove singularity qualitatively similar to that seen in Bi2212. The data taken with 21 eV photons also show significant dispersion of the higher-bindingenergy band, suggesting a band of mixed bonding and antibonding character. Dispersion relations for FY are plotted in fig. 18. While the presence of the surface feature complicates the data somewhat, presented in this way there is actually quite a large degree of consistency with the data along FX. The bonding o band appears at around 0.53 eV at F and disperses weakly downward to 0.67 eV at Y. A second band disperses upward to merge with the pure antibonding o* state in a van Hove singularity with 0.12 eV

A N G L E - R E S O L V E D P H O T O E M I S S I O N S T U D I E S OF U N T W I N N E D Y B C O

417

Y dispersion

0.1 o

0.2

A v v ~¢

0.3

o

0.4 o o o v ~;' v

O6

.... 0

.......... , 20

40

60

"m 80

OL hv=28 01, by=28 02, hv=28 03, hv=28 03, by=28 04, hv=28 04, hv=28 O4, hv=28 Ol, hv=28 01, hv=28 02, hv=28 02, hv=:28 0 k hw-21 01, hv:2~ 0 I . by:21 04, hv:=21 04, hv=2~ 04, by=21

.... 1 O0

Y Fig. 18. Energy dispersion o f quasiparticles along the FY s y m m e t r y line.

binding energy at Y. The expected chain Fermi surface crossing appears near 40% of FY, at which point a feature becomes prominent at very low binding energy. Corresponding band dispersions along XS and YS show both the bonding o and antibonding (J* bands are clearly present, with two corresponding Fermi surface crossings at 31% and 18% of XS. The fact that the higher-binding-energy bonding band is not seen in studies of twinned crystals in 02, as has been the convention in previous work, is significant, particularly as it is this state which reveals anisotropic shifts in the leading edge which mimic the gap behavior in Bi2212, as reported in previous work (Schabel et al. 1998a).

4. Fermi surface

Cuprates represent an ideal system for ARPES studies of the Fermi surface because of the direct correspondence between photoelectron emission angles and points in the Brillouin zone (BZ). This fact has been exploited in a number of experimental mappings for Bi2212, Bi2201, Y123, and NdCeCuO, revealing intriguing aspects of their electronic structure. In these experiments well-defined quasiparticle peaks are observed

418

M. SCHABELand Z.-X. SHEN

near EF and disperse with k, with Fermi level crossings corresponding to spectra in which peaks suddenly lose intensity at one point in k-space. These band-like features were initially hailed as strong confirmation of the applicability of an independentparticle model. However, the observation of substantial band-mass enhancements, satellite features, and extended saddle-point singularities casts doubt on the adequacy of bandtheoretical treatments. Investigations incorporating many-body Hamiltonians such as the t - J model predict similar behavior and provide a more satisfactory explanation for the anomalous band dispersions while retaining conceptual continuity with the insulating parent compounds (Dagotto et al. 1994). The observation of large, band-like Fermi surfaces in both electron- and hole-doped high-Te superconductors with ARPES places strong constraints on many-body theories for these systems, and has led to extensive discussion of the Luttinger sum rule and the relevance of Fermi-liquid theory to the electronic structure of the cuprates in the metallic doping regime (Campuzano et al. 1990, Dessau et al. 1993, King et al. 1993, Liu et al. 1992a,b, Olson et al. 1989b). Two important questions can be addressed by Fermi surface (FS) measurements on two-layer cuprates: (i) do two FS pockets arising from coupling of the CuO2 bilayers exist? and (ii) do "shadow bands" originating in antiferromagnetic correlations appear as echoes of the Fermi surface shifted by the antiferromagnetic wavevector, (¢c,¢c)? To adequately analyze these issues, it is essential to understand what is being measured in an ARPES experiment, how a FS is defined in a many-body correlated system which may be far from the Fermi-liquid ideal, and how that information is extracted from the experimental data. Here we briefly discuss various methods for extraction of Fermi surfaces from ARPES data (Aebi et al. 1994, Randeria et al. 1995, Santoni et al. 1991, Straub et al. 1997), along with their implementation and relative me6ts. A high-resolution mapping of the Fermi surface of Y123 was presented by Liu et al. (1992b). Due to the presence of twinning in this material, the anisotropy between the X and Y points in reciprocal space is not observable. Matrix element effects are also exceptionally pronounced in Y123, leading to strong modulation of the peak intensities with photon energy and attendant difficulties in data interpretation. The Y123 Fermi surface appears to consist of a single nested sheet centered at (g, zc) consistent with the other cuprates for which such measurements have been made, with the possibility of a second, nearly degenerate Fermi surface being less clearly resolved. The adequacy of this analysis has been questioned by some authors, and was discussed in the review by Shen and Dessau (1995). Here we emphasize the results of extensive polarization-dependent studies of the Fermi surface of untwinned Y123 (Schabel et al. 1997). In Bi2 Sr2CaCu208 + 6 (Bi2212), complications arising from superstructure have led to contradictory conclusions regarding the presence or absence of bilayer splitting, despite similar data (Dessau et al. 1993, Ding et al. 1996a), highlighting the subjectivity of conventional FS attribution. Because the predicted magnitude of the splitting in Bi2212 is much smaller than that for optimally doped Y123, the latter material is much more appropriate for such investigations. In addition, Y123 does not suffer from the Bi2212 superstructure distortion, although its FS is complicated by the presence of four distinct

ANGLE-RESOLVEDPHOTOEMISSIONSTUDIESOF UNTWINNEDYBCO

419

segments and significant kz dispersion, based on predictions of ab initio LDA calculations (Andersen et al. 1991, 1994, Pickett 1989). Also important is the fact that no complete FS mapping has been performed on an untwinned sample (Tobin et aL 1992), even though twinning renders interpretation of the data more difficult. Significant anisotropy is predicted in the band structure and FS of Y123 due to the CuO3 chains breaking the fourfold tetragonal symmetry. This leads to smearing of the spectral functions and FS of twinned crystals by superimposition of the FXSF and FYSF octants of the Brillouin zone. Interest in the possibility of directly observing strong antiferromagnetic correlation effects in the Fermi surface of the cuprates (Liu et al. 1992a) was brought to the forefront by Aebi's (1994) and Osterwalder's (1995) reports of antiferromagnetic "shadow bands" in data taken in photoelectron diffraction mode on optimally-doped Bi2212 in the normal state at room temperature. Such a dramatic manifestation of magnetic behavior in a doping and temperature regime where the antiferromagnetic correlation lengths are of the order of a planar lattice constant, at most, would have strong implications for the relevance of magnetically mediated pairing in the microscopic mechanism for superconductivity. Unfortunately, the possibility of a c(2 × 2) reconstruction in Bi2212 and other experimental difficulties hinder discrimination between such correlations and simple structural effects (Chakravarty 1995, Singh and Pickett 1995). The absence of these uncertainties in Y123, as well as a much better understanding of the details of the crystal structure, makes Y123 potentially better suited to address these questions as well. 4.1. Methods for Fermi surface determination with ARPES Conventional ARPES FS determinations infer crossings from cuts through the BZ, subjectively identifying dispersive quasiparticles "by eye" in the EDCs. While it is impossible to attribute specific, quantitative criteria to human judgment, there are a number of guidelines which are useful in identification of crossing points. First, the centroid of the quasiparticle peak can be tracked in k-space and extrapolated to its intersection with the Fermi energy, this point defining the Fermi surface boundary. Second, the total area beneath the quasiparticle peak can be monitored, with the crossing roughly corresponding to the point where this area has diminished by a factor of two. Finally, the energy position of the midpoint of the leading edge of the EDCs may be observed, with the crossing found at the point where this quantity reaches its maximum value. Depending on the particular preferences of the practitioner, one of these methods may be strictly adhered to or some synthesis of all three may be used. In either case, the subjective method is adequate for sketching out rough details of Fermi surface topology and for gaining a qualitative overview, but is neither quantitative nor unique, and cannot provide an unbiased, automatic means of extracting FS contours. An alternative technique, based on photoelectron diffraction methods (Santoni et al. 1991), has been used by Aebi et al. (1994, 1996) and Osterwalder et al. (1995, 1996) to measure the Fermi surface of Bi2212, among other materials. In this approach, the total photoemission intensity in a narrow energy window centered on EF is measured

420

M. SCHABELand Z.-X. SHEN

as a function of k by scanning over the Brillouin zone. By acquiring a single data point rather than a complete EDC for each k-vector, it is possible to densely sample the entire BZ much more rapidly and completely than is feasible in conventional EDC mode. However, there are four disadvantages to this method. First, it only provides information on the relative spectral weight near Er, but does not allow the identification of dispersive quasiparticle features or determination of the band structure. Second, matrix element effects and cross-section fluctuations may lead to sizable variation of the total photoemission intensity in different parts of the BZ, over- or underemphasizing certain regions of k-space. Third, while the technique has intuitive appeal, the precise physical meaning of the quantity being measured is not well defined. The essential assumption which is made is that the spectral weight at the Fermi level has a maximum at kv; this certainly holds in a simple Fermi-liquid model, but there is no particular reason to believe that this behavior is universal in strongly correlated systems. Fourth, in the instance of the high-To materials, or other systems where there is an extended region of k-space having a high spectral density near but below the Fermi energy, the photoelectron diffraction method is unable to resolve true FS crossings from tile near-E~ weight, and will actually include contributions from occupied states lying within the greater of the instrument resolution, a, or kT, of Ev. A third technique, which has been proposed by Randeria et al. (1995), relies on the approximate sum rule relating the spectral function to the momentum space density of states (DOS), n(k). As noted in their letter, n(k)= f ~ A ( k , co)f(o~)dco. To see how this relates to what is measured in an ARPES experiment, note that, within the sudden approximation, a valence-band ARPES spectrum may be approximated as a sum over bands of the product of a matrix element for each band with the corresponding single band spectral function: I (k, co) = ~ n Mn(k, hv) f ( co)A~(k, co). It is further conjectured that ARPES spectra may be used to directly measure n(k), through the aforementioned relationship between photoemission spectra and A(k,co). This is true within a small region of k-space, if we assume that M~(k,hv) is weakly varying and we further restrict ourselves to a single band, in which case we can integrate over co to OO obtain f~A,,I(k, co)do)= M~(k, hv)n(k), where A, is the quasiparticle bandwidth. Then the energy-integrated ARPES spectrum is proportional to the momentum-space DOS, modulated by the photon energy and k-dependence of the matrix element prefactor. Unfortunately, the situation in real, multiband materials is not nearly so simple. First, the photon energy dependence is non-trivial, and can lead to vastly different contributions from bands having different orbital character, especially in the UPS energy range. This is particularly true in structurally complex materials such as the cuprates, which have polyatomic bases with many atoms and, consequently, many overlapping bands. Second, for a global picture of the Fermi surface, the k-dependence of the matrix element term is likely to be significant on the scale of the Brillouin zone. In these cases, it is difficult to justify, a priori, straightforward application of the n(k) approach. Despite these complications, it is possible to formulate a quantitative means of directly extracting Fermi surface information from a complete set of ARPES data spanning the Brillouin zone. It is well known that, within the Fermi-liquid framework, the momentum

ANGLE-RESOLVEDPHOTOEMISSIONSTUDIESOF UNTWINNEDYBCO

421

distribution drops discontinuously at the Fermi surface, with the size of the step, Zk, being directly related to the mass renormalization. Experimental broadening from the finite angular resolution will remove this discontinuity, leading to a sigmoid function of k. In the case where multiple quasiparticles coexist, the momentum distribution will manifest several discrete drops, one for each band. If we consider the momentum space gradient of this function, [~7kn(k)[, it is clear that a peak will occur at the position of each Fermi surface crossing, corresponding to the steps in the momentum density. Under the assumption that the matrix element term is slowly varying relative to the characteristic "width" of the Fermi surface (primarily defined by the angular resolution of the photoemission spectrometer), the derivative of the photoemission spectral weight will manifest essentially identical behavior, allowing the Fermi surface to be directly extracted from the data. In cases where the matrix element variation is non-negligible, a simple first-order correction may be made using the logarithmic derivative instead: [Vkln n(k)[, thereby mapping out regions where the relative variation in n(k) is large. This can be used to correct for variations in cross-section due to polarization effects when data is acquired with synchrotron radiation or other polarized photon sources. By relying on conservation of spectral weight, the presence of weight near but below E~, will not affect the appearance of the Fermi surface determined in this way, unlike the photoelectron diffraction method, nor will the finite experimental energy resolution, since the Fermi statistics defining particle conservation are unaffected by the instrumental broadening. Similarly, conservation of spectral weight is a robust property of manybody systems, independent of the details of the interactions, making this approach equally valid for strongly correlated and non-Fermi liquid materials. An example is the Luttinger-liquid scenario, as discussed by Chakravarty et al. (1993) in which the deltafunction singularity in the gradient is replaced by a power-law divergence; while the details of n(k) are affected, the FS crossing appears in the expected region of k-space. The main disadvantage of the gradient method is the fact that numerical differentiation, which magnifies the noise already present in the data, is necessary for extraction of the Fermi surface. This is particularly problematic for data taken in normal EDC acquisition mode, necessitating a degree of numerical smoothing for good results. The required smoothing results in slightly diminished effective k-resolution relative to our actual instrumental contribution. The much denser k-space sampling of the photoelectron diffraction measurements should allow significantly improved signal-to-noise ratio in applications of the derivative method. 4.2. The Fermi surface of Y123 The Fermi surface of optimally-doped Y123, calculated using the full-potential linear muffin-tin orbital (LMTO) method is shown in figs. 19c,d for kz = 0, Jr (Andersen et al. 1991, 1994). As studies on twinned Y123 and numerous other cuprates have made clear (Dessau et al. 1993, King et al. 1993, Liu et al. 1992a), band-theory predictions are at least qualitatively correct for these materials, provide the only ab initio theoretical predictions for comparison with the experimental data, and constitute a useful framework

422

M. SCHABEL and Z.-X. SHEN F

X

F

(a)

hi

Y

F

X

F

F

F

X

F

Y

Io

F

01 (c)

X

(b)

02 (d)

Fig. 19. Fermi surfaces determined from the logarithmic gradient of the photoemission intensity at the Fermi energy. The small segments near F are artifacts due to the small effective window of ~o-integration. Panels (c) and (d) show the Fermi surfaces derived from the data (thick black and gray lines) along with Fermi surface crossings identified by conventional band dispersion measurements (open circles). Superimposed are the LMTO calculated Fermi surfaces for k~ = 0, ~ (plane states are shown in gray, chains in black).

within which to discuss possible deviations from conventional behavior (Andersen et al. 1991, 1994, 1995, Pickett 1989). The principal features of the LMTO Fermi surface are two large CuO2-plane-derived hole pockets centered on the S point, a chain-derived FS sheet running roughly parallel to FX, and a small pocket centered on S. Projection of the orbital character of the relevant bands reveals that the inner plane pocket is principally derived from the bonding pdo plane states comprised of Cu(2) 3dx2_y2, 0(2) 2px, and 0(3) 2py orbitals, while the more dispersive outer pocket derives fi'om the corresponding antibonding pdo* states. The small "stick" pocket at S arises primarily from bonding between the apical oxygen 0(4) and Ba, while the remaining pdo chain sheet weakly hybridizes with the bonding pdo* band at kz = 0 and strongly hybridizes with the antibonding pdo band at kz - ~, resulting in substantial c-axis dispersion (Andersen et al. 1994). Application of the gradient method to ARPES measurements on untwinned Y123 is presented in figs. 19a,b. Sample lifetime constraints and data acquisition time make it impossible to acquire an arbitrarily dense set of EDCs, necessitating interpolation of our data in k-space to generate a uniform mesh which is then numerically smoothed

ANGLE-RESOLVEDPHOTOEMISSIONSTUDIESOF UNTWINNEDYBCO

423

by convolution with a two-dimensional window function having a breadth 1.5 times the instrument width, and differentiated to form the quantity [Vkln n(k)[. Choice of the grids spanning the Brillouin zone was guided by previous investigations, allowing concentration of spectra in regions where there is appreciable spectral weight below E~, and sampling the regions containing unoccupied states more sparsely. The logarithmic gradient was chosen to minimize the intensity variations seen over the BZ in the raw data; other than diminished contrast, the results of application of the bare gradient are essentially indistinguishable. Examination of figs. 19a and 19b reveals the expected large hole pocket centered on S, with significantly narrower Fermi sections than observed in the photoemission intensity maps. In contrast with the photoemission intensity maps discussed above, the segments resolved with the gradient are comparable in width to the instrumental k-resolution defined by our angular aperture (broadened by smoothing). It was verified from the band dispersions that the parallel segments near the F point result from the finite energy window, with a quasiparticle dispersing up from higher binding energy passing first through the lower edge of the energy window, then crossing the Fermi surface, as discussed above. The remaining portions are sections of the true Fermi surface, although segments in the vicinity of the surface peak at Y are aliased by the spurious spectral weight of this feature and are not representative of the bulk Fermi surface. In both orientations we can clearly resolve the strongly nested inner (bonding) plane FS pocket, with weak indications of the outer (antibonding) pocket visible primarily near X and Y. While the derivative data are too noisy to unambiguously reveal both crossings along FS, detailed analysis of dispersion relations along this line clearly demonstrates their presence (Schabel et al. 1998a). The chain band is weakly apparent, though it presumably overlaps to a large degree with the antibonding plane segments around X. Figures 19c,d show subjective Fermi surfaces, determined from the plots in figs. 19a,b, as thick lines. Superimposed are LMTO Fermi surface contours for kz = 0, :v, with the plane bands in grey and chains in black. Experimental FS crossings determined from band dispersions using the conventional method are indicated by the open circles (Schabel et al. 1998a). The expected bifurcated saddle points in the antibonding Fermi surface may be resolved in segments paralleling FY, indicated by the thick gray lines. The ability to make quantitative statements about this region of the Brillouin zone is, however, compromised by the large contribution of the surface peak near Y. A complimentary bifurcation of the saddle point in the FX-direction, if present, is not well resolved, in contrast with LDA predictions. This is likely a consequence of strong c-axis dispersion distributing spectral weight over a larger region of k-space for the dispersive pdo* FS sheets in this part of the Brillouin zone. In 02, where E is parallel to the CuO3 chains, we note that even using the logarithmic gradient technique the emission from the feature near Y is by far the most prominent contribution, while the relative contributions are comparable in Ol. To identify Fermi surface segments and crossing points more precisely, it is possible to examine I(EF) contours along lines paralleling both FY and FX for various values of kx and ICy, respectively. In this approach, which is complementary to band-structure identification, with a peak being identified in k for fixed ~o=EF, Fermi surface crossings

424

M. SCHABELand Z.-X. SHEN

1.0 J

\

X

.J\\

.j-~\

0.5 0.5

ky

1

0.5

ky

1

Fig. 20. Normalized cuts through the data used in fig. 19, parallel to FY for k.~ ranging between 0.5 and 1.0. Two peaks corresponding to two separate Fermi surface crossings appeal"in the data in both sample geometries.

should appear as peaks in the intensity as a function o f k. Figure 20 shows normalized Fermi-level intensity curves for cuts parallel to FY for kx between 0.5 and 1.0, corresponding to cross-sectional slices o f the data. The data in both plots have a double peaked structure extending over a large fraction o f the FX line, with a smaller feature near the Icy = 0 line and a more prominent one around ky = 0.2, which merge into a single broader peak nearer to F in O1, but remain distinct in 02. An important technical detail which arises when performing such an analysis is the occurrence o f spurious peaks when the line along which I(E~.) is being interpolated lies parallel (or nearly so) to a segment of the Fermi surface. In this case, the actual Fermi "crossing" will be extremely broad, and noise in the data may give rise to features which do not correspond to the true Fermi surface. For this reason, data from these regions must be regarded as unreliable. 4.3. Shadow bands in Y123 The shadow bands observed in the Fermi surface o f Bi2212 by Aebi et al. (Aebi et al. 1994, Osterwalder et al. 1995) seen in normal-state measurements o f the photoemission intensity at room temperature, were attributed to echoes o f the bonding FS shifted by (st, Jr) arising from strong antiferromagnetic correlations. This interpretation has been controversial; the appearance o f these features in their experiment is particularly surprising because the data were taken on optimally doped, metallic Bi2212 at 300K, which has an extremely short antiferromagnetic correlation length on the order o f a planar lattice constant. A possible alternative explanation o f this effect is a weak c(2×2)

ANGLE-RESOLVED P H O T O E M I S S I O N STUDIES OF U N T W I N N E D Y B C O

425

structural distortion which would mimic the effects of static antiferromagnetic order. The data on untwinned Y123, which were taken at low temperature (20 K), provide an independent test of this hypothesis on a material known to be free of such structural distortions. If static antiferromagnetic long-range order were present in Y123, we would expect the true LMTO FS to be mirrored by shifted echoes of the folding from the antiferromagnetic potential. One should be able to directly observe these echoes in the photoemission data experimentally, with the most prominent features of the unperturbed Fermi surface presumably having the strongest shadows as well. Examination of the experimental Fermi surfaces in fig. 19 does not reveal any indication of such a folding in any of the Y123 data. in addition, close inspection of EDCs along the FS line shows no evidence for the presence of a peak shifted by (g~,5~) 2 2 corresponding to the main 1 1 Fermi surface crossing near (5zc,5~c) (Schabel et al. 1997). These data suggest that the shadow bands seen in Bi2212 are structural in origin, and do not arise from the vestigial antiferromagnetic order in this material, and are consistent with earlier investigations searching for closed pockets along FS in twinned Y123 (Liu et al. 1992a). 4.4. Evidence for bilayer splitting Bilayer splitting, which arises from bonding-antibonding pairing of interplane wavefunctions, is predicted to be approximately 0.75eV at F in Y123, substantially larger than that expected in Bi2212. Some theoretical predictions, principally based on the short c-axis mean free paths in normal-state transport data, have argued that the conductivity data fundamentally contradict band-like behavior along this axis, thus precluding the possibility of this splitting (Anderson 1992). In particular, the interlayer tunneling mechanism of high-temperature superconductivity relies on incoherence of the inter-bilayer transport in the normal state. For this reason, it is important to critically evaluate the evidence for this splitting in Y123 and discuss its significance. The hallmark of bilayer splitting is the appearance of two nondegenerate bands arising from the CuO2 planes, while a degeneracy of these bands might be, but is not necessarily, an indication of non-Fermi liquid behavior. While the data are normally taken in the superconducting state, the vastly different energy scales for the expected splitting and the superconductivity-induced modifications in the electronic structure make it unlikely that this is a relevant consideration. As discussed above, the data clearly reveal two plane-derived quasiparticle features which have binding energies of 0.22 eV and 0.53 eV at F for hv = 28 eV, with the higherenergy peak showing at least 0.12 eV dispersion along the c-axis. In contrast, there is no apparent c-axis dispersion of the low-binding-energy feature in our data. These features disperse weakly upward toward X, appearing at 0.13 eV and 0.43 eV respectively, and exhibit Fermi surface crossing behavior agreeing well with qualitative aspects of LDA calculations. On the basis of this evidence, it appears that these features are in fact the expected bilayer split plane states, with the magnitude of the splitting diminished by roughly a factor of two from its calculated value, presumably due to correlation effects.

426

M. SCHABELand Z.-X. SHEN

The apparent observation of bilayer splitting and c-axis dispersion in Y123 raises some intriguing questions regarding the qualitative disagreement between high-energy measurements such as photoemission and low-energy measurements such as transport. In most of the cuprates, the c-axis mean free path for electron transport is shorter than the lattice constant in that direction, making it plausible to argue that the transport in this direction is incoherent (Anderson 1992). However, this conclusion, based on the results of transport measurements at very small energy scales, does not necessarily imply that high-energy spectroscopies such as photoemission should not see fundamentally different effects (Laughlin 1997). In fact, the observation of bilayer splitting and c-axis dispersion in Y123 points to the need for a conceptual distinction between the two very different energy scales probed by these different techniques. Such a duality in which band-like dispersive quasiparticles are seen in photoemission measurements of materials which are clearly insulating in transport has been seen in a number of other systems, including NiO, Sr2CuO2C12, and other Mott insulators.

5. Superconducting-state electronic structure

Observation of a superconducting gap for the first time in any material using photoemission was a remarkable accomplishment, and paved the way for the intense research which focussed on determination of the momentum-space dependence of the superconducting order parameter (Imer et al. 1989). Unfortunately, aside from isolated and largely irreproducible results on Y123 (Schroeder et al. 1993), Bi2212 was the only cuprate for which a gap could be reliably and reproducibly measured. Because of the absence of consistent data on other cuprates, a vast amount of work has focussed on Bi2212, leading to significant advances in our understanding of the superconducting state and sharpening the debate on the origin and character of the microscopic pairing mechanism. Initial observations of the opening of a gap in the near-EF states in Bi2212 were made in angle-integrated mode, demonstrating a shift in the leading edge of 20-30 meV and a pile-up of states similar to that expected from BCS theory (Chang et al. 1989, lmer et al. 1989). Early angle-resolved studies revealed an even more dramatic change in the DOS, finding an isotropic gap of 2 4 i 5 meV (Olson et al. 1990b). This evidence for a symmetric s-wave pairing mechanism contrasted strongly with numerous other experiments which clearly indicated the presence of an anisotropic gap; subsequent ARPES work by Wells et al. (1992) revealed for the first time a direct observation of k-dependence of the gap. The sensitivity of gap measurements to scattering and disorder and the high energy resolution required to observe the appearance of a gap make these studies particularly demanding in terms of sample quality and surface preparation, with imperfections leading to homogenization of the angular dependence and an apparently isotropic gap. Extensive studies by Shen et al. (1993), Shen (1994) and Ding et al. (1996a,b), have examined the question of gap anisotropy in Bi2212 in detail, finding strong evidence for dx2 / symmetry in numerous samples. Due to their lack of phase sensitivity, ARPES measurements are incapable of rigorously determining the k-dependence

ANGLE-RESOLVEDPHOTOEMISSIONSTUDIESOF UNTWINNEDYBCO

427

of the superconducting order parameter; however, it is possible to set an upper bound on the magnitude of the gap along a hypothetical node line. Comparison of the momentum-space dependence observed with that expected for d-wave symmetry, ( A ( k ) = A o ( c o s k x a - c o s k y a ) , demonstrates a strong linear correlation between the experimental data on a number of samples and the simple d-wave model. Alternate predictions for the order-parameter symmetry, such as the anisotropic s-wave, exist (Chakravarty et al. 1993), but a growing body of evidence from a number of techniques strongly favors the d-wave scenario. At present, the emerging consensus based on both ARPES and other phase-sensitive measurements (Ott and Brawner 1995, Tsuei et al. 1995, van Harlingen 1995, Wollman et al. 1995), appears to strongly favor pairing mechanisms which lead to d-wave symmetry. Recently, several elegant experiments on Y123 have provided strong evidence for a phase change under 90 ° rotation, also with the d-wave hypothesis (Ott and Brawner 1995, Tsuei et al. 1995, Wollman et al. 1995). Angle-resolved photoemission, although only sensitive to the magnitude of the order parameter, is nevertheless able to place strong upper bounds on the maximum gap magnitude at the node line, and has the unique capability of directly measuring the k-dependence of this quantity. Unfortunately, previous attempts to measure the gap in Y123 have failed. This absence is particularly puzzling in light of numerous results obtained with other experimental techniques, including Josephson tunneling (van Harlingen 1995), microwave absorption (Bonn and Hardy 1996), and scanning SQUID microscopy (Kirtley et al. 1995), which clearly reveal the appearance of a superconducting gap of comparable magnitude and anisotropy to that seen in Bi2212 (Shen and Dessau 1995). The absence of a gap is also inconsistent with scanning tunneling microscopy studies (Edwards et al. 1992), which are even more surface sensitive than ARPES. In fig. 21 we present photoemission spectra from sample X III taken along six distinct cuts through the two-dimensional BZ, normalized to the weight above EF arising from higher-order photons. Each panel clearly reveals a Fermi surface crossing, with a schematic representation of the points in the BZ corresponding to the spectra shown beneath. A crossing is identified by an abrupt loss of intensity in the quasiparticle feature as it moves above EF. The importance of accurately determining the precise k-space location of the crossing has been extensively discussed in the context of gap studies on Bi2212 (Shen et al. 1993). Crossings are identified using two criteria: (i) the point where the integrated intensity in the quasiparticle is reduced by half, and (ii) the point where the midpoint of the leading edge near E~ reaches the farthest forward (i.e. the lowest binding energy). Figure 22 compares the leading edge regions of spectra taken at the FS crossing points; only four of the six crossings are shown for clarity. The Fermi edge spectrum from a Au reference sample is indicated by the thick line. A monotonic shift of the leading edge is seen as we move along the FS from FS toward XS, with the clear appearance of a gap in the spectra near X. While this behavior is similar to the gap phenomena seen in Bi2212, there are a number of important differences. In particular, we fail to observe the sharp peak which appears prominently near 1VI (equivalent to X in Y123) in

428

M. SCHABEL and Z.-X. SHEN

I

© ctz

L

© ©

0.00

0.50

0 oo

o.5o

0.00

o.5o

o.oo

0.50

0 oo

o ~0

0.00

Binding Energy (eV) Y

S

•

Y

'

"" . . . . . .

X

.i_..

"-. .....

S

.....

X

Fig. 2 l. Photoemission spectra along various cuts through the Brillouin zone, clearly revealing a Fermi surface crossing. The lower portion schematically illustrates the points in k-space corresponding the each spectrum. A second Fermi surface sheet is indicated by the hatched line.

the superconducting state. Also, spectra along FS exhibit qualitatively different crossing behavior when compared with those along XS, with the former showing a pronounced sharpening while the latter diminish in intensity without significant changes in the [ineshape. Finally the leading edges of the spectra nearest the FS line appear to lie significantly above the Fermi level as determined by our reference sample. This effect is also seen in Bi2212, but it is markedly more pronounced in our Y123 data. These observations are likely to be closely related, and may be understood by considering the interplay between the underlying spectral function and the finite instrument resolution. The absence of a sharp peak at the XS crossing is a source of concern, particularly since it is such a prominent feature in the superconducting state of Bi2212. However, tbe presence of a gap of unknown origin in the excitation spectrum is clearly indicated by the data, and has been reproduced on four separate samples. Unlike Bi2212 in which the bands near (~c,0) are quite flat and remain near the Fermi level over an extended region of k-space, we observe significant dispersion (-0.5 eV) near the Fermi surface crossing along XS, although an additional flatter band may also be seen near the X point. This large dispersion, in conjunction with decreased momentum resolution at the higher photon energies used on Y123, can lead to broadening of the spectra. We can make

ANGLE-RESOLVED PHOTOEMISSION STUDIES OF UNTWINNED YBCO r

429

7-

\o

\'~'~ \ ~'~\

=o

Y

S ~\~'['~\\

L

©

t~EI0.10

0.05

0.00

-0.05

Binding Energy (eV)

0.10

Fig. 22. The near-El~ region of spectra at the Fermi surface crossing are superimposed, revealing the presence of a monotonic shift of the leading edge to higher binding energies from the FS crossing to the XS crossing. A reference spectrum for a Au sample is shown by a thick line for comparison. Spectra are normalized for purposes of visualization.

a simple quantitative estimate of the momentum-related broadening by comparing the angular resolution of approximately 10% of the FX distance with the observed dispersion to find a total contribution of N150meV, significantly larger than the experimental energy resolution. In addition, there is some evidence to substantiate the LDA prediction of two distinct plane-derived Fermi surface sheets. The presence of such a bifurcation could lead to a strong overlap of two quasiparticle features in the photoemission spectra, obscuring any sharp features. With these caveats in mind, we believe that it is most reasonable to attribute the leading-edge shifts in the data to modifications due to the presence of a superconducting gap. This interpretation leads to a self-consistent picture which agrees with photoemission measurements of the superconducting gap on Bi2212 and with other experimental results on Y123. In this picture, the shifts of the leading edge along the Fermi surface will reflect, to first order, the relative variation in the magnitude of the superconducting gap at different points along the FS. The measured leading edge shifts are presented in fig. 23. Due to the experimental uncertainty in determining the absolute magnitude of the gap along FS, all shifts are measured relative to the position of the leading edge at this crossing. The error bars in the figure represent the uncertainty in determination of the shift due to lineshape issues discussed above; errors due to uncertainty in the fitting parameters or in the reference EF position are substantially smaller. Comparison of the relative shift with the characteristic k-space dependence of a d-wave order parameter, Icos kx- cos ky I/2, reveals clear linear dependence for three of the four samples (X III, X VII, and X III 2), while the data from sample X V shows an extended node region, with the gap rapidly

430

M. SCHABEL 7o

I

'

'

'

I

'

'

'

I

'

'

'

+

"

'

+

i

'

and Z.-X. SHEN

'

"

'

-sampleX III

'

'

i

•

.

i

sample X V

25

20 v L5

O3 5 ¢) ;>

0 I~r i

d ,

000

--~ 30

I

r i

r I

0,20 '

'

'

r

i

T ,

0+40 '

'

'

t

I r

i

,

0.60 *

'

•

E

I I

i

0.80 '

'

'

i

' 0

, 1.00

'

'

i

i n

0.00

'

.isample X VII

(b)

I

i

I

0.20

'

'

'

,

,

,

0,40

i

,

,

,

,

0.60

i

,

,

,

,

i

i

i

i

,

,

,

l

,

,

0.80

i

,

,

t

,

,

1.oo

sa

25

'~ 15

5

5

0 000

0 J

l

r

I

I

0.20

0.40

0.60

0.80

l.O0

i

0.00

t

(d) ,

0.20

Icos k x - cos kyl/2 Fig. 23.

Measured

crossings

along

shifts the

and X IIl 2 exhibit extended

Fermi

(meV)

strong

linear

region

of zero

plotted

correlations,

,

l

0.60

,

0.80

LO0

Icos k x - cos kyl/2

in the leading

surface,

t

0.40

edges

relative

of the photoemission to the

consistent

shift in sample

quantity

with

X V may

spectra

Icoskx

an order parameter be indicative

for four

coskvl/2. having

of "dirty"

samples,

Samples

x

taken

III, x

dxz y2 s y m m e t r y .

d-wave

at

vii, The

behavior.

growing only near the zone boundary. This behavior may be representative of a "dirty" d-wave material, and is consistent with our observations that the quasiparticle feature was substantially less distinct and the cleave visually inferior to those in other samples. Extrapolation of the linear region to the zone boundary provides values for 2Ao/kT in the range of 5.4-7.6. Our inability to compare spectra above and below Tc in Y123 makes careful analysis of lineshapes crucial. Simple numerical simulations of photoemission spectra can provide insight into the behavior seen in fig. 22, the details of which have been presented by Schabel et al. (1998b). Relying on the spectral function interpretation of the photoemission process, we can model a quasiparticle dispersing through EF with a simple Lorentzian peak, cut by a Fermi function and convolved with a Gaussian corresponding to the experimental energy resolution. In order for a spectrum to exhibit a large shift of the leading edge above EF, it is necessary for the underlying quasiparticle to be narrow and lie near the Fermi level, in which case the instrumental broadening will lead to a forward shift on the order o f the Gaussian standard deviation. From the figure, we see a shift of approximately 15 meV, which is of the magnitude expected given the experimental

ANGLE-RESOLVED PHOTOEMISSION STUDIES OF UNTWINNED YBCO

431

resolution. This implies that the peak at the FS crossing must have a small or zero gap. It is also important to note that the appearance of a gap in our photoemission spectra, as indicated by a shift of the leading edge below the Fermi level of the Au reference, cannot be explained without invoking a real gap. The failure of previous studies to observe a superconducting gap in Y123 (see Veal and Gu 1994 for a review of earlier work), despite six years of effort, requires some explanation. Two critical elements of these experiments differ from the earlier attempts, enabling observation of the presence of leading edge shifts. First, untwinned samples, in which the CuO3 chain axes are uniformly oriented, are used. Because twinning mixes the normally inequivalent X and Y points, significant mixing of the spectral function occurs throughout the Brillouin zone. Second, the photoemission spectra in three distinct sample orientations relative to the plane of photon polarization are examined, which enables identification of a previously unseen dispersive feature which is strongly enhanced in only one geometry. Also, the large dispersion of this band makes sample alignment critical, unlike the case of Bi2212. It is also important to note that, as in previous studies, we do not observe gap-like shifts in the leading edges of spectra taken near the X/Y points; our measurements correspond to the innermost Fermi surface sheet centered on the S point. This absence of a gap remains a puzzle, perhaps related to the strong hybridization of chain and antibonding plane states in conjunction with electronic structure effects from the chain-terminated surface.

6. Summary In summary, extensive studies of the dispersive quasiparticle states in optimally doped untwinned single crystals of Y123 using polarization-dependent angle-resolved photoelectron spectroscopy have illuminated a number of points regarding the electronic structure of this material. Based on analysis of the Ba 5p core level and inference from other surface-sensitive probes, the observed surface-to-bulk intensity ratio can be reproduced with a simple model incorporating a mixed surface termination of BaO and CuO3 regions. Furthermore, with an electron attenuation length in this material of approximately 5 A, the bulk of the photoemission intensity arises from the surface and first subsurface CuO2 planes, with significant attenuation of the signal from the bulk chains. Extensive arguments, based on photon polarization dependence, photon energy dependence, intensity variations between twinned, untwinned, and Y124 crystals, and oxygen, Co, and Pr doping results, favor attribution of the intense, narrow feature found near the Y point to a surface-related chain state. This feature, which appears at both X and Y in twinned crystals, is extrinsic to the bulk electronic states of Y123, and contaminates data in the crucial near-El, region in crystals which are not exceptionally well detwinned. The measured band dispersions clearly reveal two plane-derived quasiparticle states, one with a binding energy of 0.22 eV at F, and the other with 0.53 eV, which are associated with the antibonding and bonding bilayer bands. The former disperses upward to form a van Hove singularity with a binding energy of 0.13 eV at X, while the latter is weakly

432

M. SCHABEL and Z.-X. SHEN

dispersive along FX, and disperses rapidly upward along XS. While no measurable c-axis dispersion is seen in the antibonding band by changing the photon energy, the bonding band shifts upward by 0.11 eV when h v is changed from 2 8 e V to 21eV. Observation o f two distinct plane states and c-axis dispersion provides clear evidence for bilayer splitting in Y123, consistent with the simplest band-theoretical arguments. Quantitative Fermi surface determination based on a momentum-space DOS analysis reveals a strongly nested inner Fermi surface pocket centered on S attributed to the CuO2 bonding pdo bands. The bilayer splitting o f the plane bands, which is clearly resolved in polarization-dependent energy-distribution curves appears as a more weakly resolved outer sheet in the k-space data. Vestiges o f the principal chain sheet are also weakly apparent in some o f the data, but are significantly suppressed, consistent with a strongly modified surface chain termination. The data show no sign o f a "stick" Fermi surface centered at S, consistent with previous measurements but disagreeing with LMTO calculations. Evidence for antiferromagnetic shadow Fermi surfaces appears to be lacking in optimally doped Y123, supporting a structural origin for such features seen in Bi2212. Finally, the presence o f leading-edge shifts with a k-dependence characteristic o f pure d-wave order parameter untwinned Y123 has been observed. These shifts cannot be conclusively associated with the onset o f the superconducting transition due to the surface instability o f this material above Tc, but are consistent with a highly anisotropic superconducting gap o f d-wave form. The lack o f phase information makes it difficult to exclude the possibility o f symmetries such as d + id or highly anisotropic s-wave mimicking the d-wave form. References Abrikosov,A., LC. Campuzano and K. Gofron, 1993, Physica C 214, 73. Aebi, R, J. Osterwalder, E Schwaller, L. Schlapbach, M. Shimoda, T. Mochiku and K. Kadowaki, 1994, Phys. Rev. Lett. 72, 2757. Aebi, R, T. Kreutz, J. Osterwalder, R. Fasel, R Schwaller and L. Schlapbach, 1996, Phys. Rev. Lett. 76, 1150. Alexandrov, A.S., and N.E Mott, 1995, Polarons and Bipolarons (World Scientific, Singapore). Allen, J., 1985, J. Magn. Magn. Mater. 47-48, 168. Allen, J., 1992, in: Synchrotron Radiation Research: Advances in Surface and Interface Science, Vol. 1: Techniques, ed. R. Bachrach (Plenum, New York) ch. 6. Andersen, O.K., A.I. Liechtenstein, C.O. Rodliguez, l.l. Mazin, O. Jepsen, VR Antropov, O. Gunnarsson and S. Gopalan, 1991, Physica C 185, 147. Andersen, O.K., O. Jepsen, A.L Liechtenstein and I.I. Mazin, 1994, Phys. Rev. B 49, 4145.

Andersen, O.K., A.I. Liechtenstein, O. Jepsen and E Paulsen, 1995, J. Phys. Chem. Solids 56, 1573. Anderson, P.W, 1987, Science 235, 1196. Anderson, RW, 1990, Phys. Rev. Lett. 64, 1839. Anderson, P.W, 1992, Science 256, 1526. Arko, A.J., R. List, R. Bartlett, S.-W. Cheong, Z. Fisk, J.D. Thompson, C. Olson, A.-B. Yang, R. Lin, G. Gu, B.V~ Veal, J. Lin, A. Paulikas and K. Vandervoort, I989, Phys. Rev. B 40, 2268. Bardeen, J., L.N. Cooper and J.R. Schrieffer, 1957, Plays. Rev. 108, 1175. Bednorz, J.G., and K.A. Mfiller, 1986, Z. Phys. B 64, 189. Berglund, C., and WE. Spicer, 1964, Phys. Rev. A 136, 1030. Bonn, D.A., and W.N. Hardy, 1996, in: Physical Properties of High-Temperatm'e Superconductors, Vol. V,ed. D. Ginsberg (World Scientific,Singapore) ch. 9. Bulut, N., and D.J. Scalapino, 1995, in: Proc. Con£ on

ANGLE-RESOLVED PHOTOEMISSION STUDIES OF UNTWINNED YBCO Spectroscopies in Novel Superconductors, Stanford 1995, J. Phys. Chem. Solids 56, 1597. Callaway, J., mad N. March, 1984, in: Density Functional Methods: Theory and Applications, Vol. 38 of Solid State Physics, ed. H. Ehrem'eich (Academic Press, New York) p. 135. Campuzano, J.C., G. Jennings, M. Faiz, L. Beaulaigue, B. Veal, J. Liu, A. Paulikas, K. Vandervoort, H. Claus, R. List, A.J. Arko and R. Bartlett, 1990, Phys. Rev. Lett. 64, 2308. Campuzano, J.C., G. Jennings, A.J. Arko, R. List, B. Veal and R. Benedek, 1991, J. Phys. Chem. Solids 52, 1411. Campuzano, J.C., K. Gofron, R. Liu, H. Ding, B. Veal and B. Dabrowski, 1992, J. Phys. Chem. Solids 53, 1577. Cardona, M., and L. Ley, eds, 1978a, Photoemission in Solids, Vol. I (Springer, Berlin). Cardona, M., and L. Ley, eds, 1978b, Photoemission in Solids, Vol. II (Springcr, Berlin). Chakravarty, S., 1995, Phys. Rev. Lett. 74, 1885. Chakravarty, S., A. Sudbo, RV~ Anderson and S.R Strong, 1993, Science 261, 337. Chang, Y., M. Tang, R. Zanoni, M. Onellion, R. Joynt, D. Huber, G. Margaritondo, R Morris, W. Bonner, J.M. Tarascon and N. Stoffel, 1989, Phys. Rev. B 39, 4740. Chu, C.W., L. Gao, E Chen, Z.J. Huang, R.L. Meng and Y. Xue, 1993, Nature 365, 323. Claessen, R., R. Anderson, J. Allen, C. Olson, C. Janowitz, V~ Ellis, S. Harm, M. Kalning, R. Manzke and M. Skibowski, 1992, Phys. Rev. Lett. 69, 808. Claessen, R., R. Anderson, G.-H. Gweon, J. Allen, W. Ellis, C. Janowitz, C. Olson, Z.-X. Shen, V. Eyert, M. Skibowski, K. Friemelt, E. Bucher and S. Htifner, 1996, Plays. Rev. B 54, 2453. Claessen, R., T. Stranb, R Steiner, S. Hfifner, V Eyert, R. Anderson, J. Allen, C. Janowitz and C. Olson, 1997, Physica B 230-232, 294. Cox, R, 1992, Transition Metal Oxides (Clarendon, Oxford). Dagotto, E., 1994, Rev. Mod. Phys. 66, 763. Dagotto, E., and T.M. Rice, 1996, Science 271, 618. Dagotto, E., J. Riera, Y. Chen, A. Moreo, A. Nazarenko, E Alcaraz and E Ortolani, 1994, Phys. Rev. B 49, 3548. Davis, L., 1982, Phys. Rev. B 25, 2911. Dessau, D.S., Z.-X. Shen, D.M. King, D.S. Marshall, L.W. Lombardo, R Dickinson, A.G. Loese~; J. DiCarlo, C.-H. Park, A. Kapitulnik and W.E. Spicer, 1993, Phys. Rev. Lett. 71, 2781.

433

Ding, H., A. Bellman, J.C. Campuzano, M. Randeria, M.R. Norman, T. Yokoya, T. Takahashi, H. Katayama-Yoshida, T. Mochiku, K. Kadowaki, G. Jennings and G. Brivio, 1996a, Phys. Rev. Lett. 76, 1533. Ding, H., M.R. Norman, J.C. Campuzano, M. Randeria, A. Bellman, T. Yokoya, T. Takahashi, "12Mochiku and K. Kadowaki, 1996b, Phys. Rev. B 54, 9678. Edwards, H.L., J.T. Markert and A.L. de Lozanne, 1992, Phys. Rev. Lett. 69, 2967. Edwards, H.L., A.L. Barr, J.T. Markert and A.L. de Lozannc, 1994, Phys. Rev. Lett. 73, 1154. Emery, V.J., and S.A. Kivelson, 1993, Physica C 209, 597. Emery, V.J., and S.A. Kivelson, 1995, Nature 374, 434. Fehrenbacher, R., and T.M. Rice, 1993, Phys. Rev. Lett. 70, 3471. Fisk, Z., and J. Sarrao, 1995, J. Phys. Chem. Solids 56, 1891. Fuggle, J.C., and J.E. Inglesfield, eds, 1992, Unoccupied Electronic States: Fundamentals for XANES, EELS, IPS and BIS (Springer, Berlin). Fujimori, A., E. Takayama-Muromachi, Y. Uchida and B. Okai, 1987, Phys. Rev. B 35, 8814. Fukuyama0 H., 1988, Solid State Phys. 23, 406. Gofron, K., J.C. Campuzano, H. Ding, C. Gu, R. Liu, B. Dabrowski, B. Veal, W. Cramer and G. Jermings, 1993, J. Phys. Chem. Solids 54, 1193. Gofron, K., J.C. Campuzano, A. Abrikosov, M. Lindroos, A. Bansil, H. Ding, D. Koelling and B. Dabrowski, 1994, Phys. Rev. Lett. 73, 3302. Gu, C., B. Veal, R. Liu, H. Ding, A. Paulikas, J.C. Campuzano, R Kostid, R. Wheelm; H. Zhang, C. Olson, X. Wu and D. Lynch, 1993, J. Phys. Chem. Solids 54, 1177. Guo, G., and W. Temmerman, 1990, Phys. Rev. B 41, 6372. Halbritter, J., R Walk, H.-J. Mathes, W. Aarnink and I. Apfelstedt, I988, Am. Inst. Phys. Cone Proc. 182, 208. Htifner, S., 1996, Photoelectron Spectroscopy: Principles and Applications, Vol. 82 of Springer Series in Solid-State Sciences (Springer, Berlin). Imer, J.-M., E Patthey, B. Dardel, W.-D. Scheider, Y. Baer, 5(. Petroff and A. Zettl, 1989, Phys. Rev. Lett. 62, 336. Jones, R., and O. Gunnarsson, 1989, Rev. Mod. Phys. 61, 689.

434

M. SCHABEL and Z.-X. SHEN

Kampf, A.R, and J.R. Sehrieflbr, 1990, Phys. Rev. B 42, 7967. Kevan, S., ed., 1992, Angle-resolved Photoemission: Theory and Current Applications (Elsevier, Amsterdam). Khomskii, D., 1993, J. Superconductivity 6, 69. Kim, C., A. Matsuura, Z.-X. Shen, N. Motoyama, H. Eisaki, S. Uchida, 32 Tohyama and S. Maekawa, 1996, Phys. Rev. Lett. 77, 4054. King, D.M., Z.-X. Shen, D.S. Dessau, B.O. Wells, W.E. Spicer, A.J. Arko, D.S. Marshall, J. DiCarlo, A.G. Loeser, C.-H. Park, E. Ratner, J. Peng, Z. Li and R.L. Greene, 1993, Phys. Rev. Lett. 70, 3159. Kirtley, J.R., C.C. Tsuei, J.Z. Sun, C.C. Chi, L.S. YuJahnes, A. Gupta, M. Rupp and M.B. Ketchen, 1995, Nature 373, 225. Laughlin, R., 1988, Science 242, 525. Laughlin, R., 1997, private communication. Levi, B., 1993, Phys. Today 46, 17. Liang, J., X. Xu, S. Xie, G. Rao, X. Shao and Z. Duan, 1987, Z. Phys. B 69, 137. Liechtenstein, A.I., and l.I. Mazin, 1995, Phys. Rev, Lett. 74, 1000. Lin, C., W. Zhun, W Liang, E. Sch6nherr and H. Bender, 1992, Pbysica C 195, 291. List, R., A.J. Arko, Z. Fisk, S.-W. Cheong, S. Conradson, J.D. Thompson, C. Pierce, D. Peterson, R. Bartlett, N. Shinto, J. Schirber, B.W. Veal, A. Paulikas and J.C. Campuzano, 1988, Phys. Rev. B 38, 11966. List, R., A.J. Arko, R. Bartlett, C, Olson, A.-B. Yang, R. Liu, C. Gu, B. Veal, YiC. Chang, E-Z. Jiang, K. Vandervoort, A. Paulikas and J.C. Campuzano, 1989, Physica C 159, 439. Liu, L., R. Anderson and J. Allen, 1991, J. Plays. Chem. Solids 52, 1473. Liu, R., B. Veal, A. Paulikas, J. Downey, R Kostid, S. Fleshier, U. Welp, C. Olson, X. Wu, A.J. Arko and J. Joyce, 1992a, Phys. Rev. B 46, 11056. Liu, R., B. Veal, A. Paulikas, J. Downey, H. Shi, C. Olson, C. Gu, A.J. Arko and J. Joyce, 1992b, Phys. Rev. B 45, 5614. Loeser, A.G., Z.-X. Shen, D.S. Dessau and W.E. Spicer, 1994, J. Electron Spectrosc. Relat. Phenom. 66, 359. Maeda, H., Y. Tanaka, M. Fnkutomi and T. Asano, 1988, Jpn. J. Appl. Phys. 27, 574. Markiewicz, R., 1991, Int. J. Mod. Phys. B 5, 2037. MilIis, A.J., H. Monien and D. Pines, 1990, Phys. Rev. B 42, 167.

Mitzi, D.B., L.W. Lombardo, A. Kapitulnik, S. Laderman and R. Jacowitz, 1990, Phys. Rev. B 41, 6564. Monthoux, R, P. Balatsky and D. Pines, 1991, Phys. Rev. Lett. 67, 3448. Nagaosa, N., and RA. Lee, 1991, Phys. Rev. B 43, 1233. Okai, B, M. Kosuge, H. Nozaki, K. Takahashi and M. Ohta, 1988, Jpn. J. Appl. Phys. 27, L41. Olson, C., R. Liu, D. Lynch, B. Veal, Y.C. Chang, RZ. Jiang, J. Liu, A. Paulikas, A.J. Arko and R. List, 1989a, Physica C 162, 1697. Olson, C., R. Liu, A.-B. Yang, D. Lynch, A.J. Arko, R. List, B. Veal, Y,C. Chang, E-Z. Jiang and A. Paulikas, 1989b, Science 245, 731. Olson, C., R. Liu, D. Lynch, R. List, A.J. Arko, B. Veal, Y.C. Chang, R-Z. Jiang and A. Paulikas, 1990a, Phys. Rev. B 42, 381. Olson, C., R. Liu, A.-B. Yang, D. Lynch, A.J. Arko, R. List, B. Veal, "LC. Chang, E-Z. Jiang and A. Paulikas, 1990b, Science 245, 731. Osterwalder, J., E Aebi, R Schwaller, L. Schlapbach, M. Shimoda, T. Mochiku and K. Kadowaki, 1995, Appl. Phys. A 60, 247. Osterwalder, J., 32 Greber, E Aebi, R. Fasel and L. Schlapbach, 1996, Phys. Rev. B 53, 10209. Ott, H.R., and D.A. Brawner, 1995, in: Proc. Conf. on Spectroscopies in Novel Superconductors, Stanford 1995. J. Phys. Chem. Solids 56, 1783. Pickett, W.E., 1989, Rev. Mod. Phys. 61, 433. Pickett, V~E., 1995, private communication. Radousky, H., 1992, J. Mater. Res. 7, 1917. Randeria, M., H. Ding, J.C. Campuzano, A. Bellman, G. Jennings, 32 Yokoya, T. Takahashi, H. KatayamaYoshida, 32 Mochiku and K. Kadowaki, 1995, Phys. Rev. Lett. 74, 4951. Rice, T.M., S. Gopalan and M. Sigrist, 1994a, Phys. Rev. B 199, 378. Rice, T.M., S. Gopalan, M. Sigrist and EC. Zhang, 1994b, J. Low Temp. Phys. 95, 299. Satje, E.K.H., A.S. Alexandrov and W. Liang, eds, 1995, Polarons and Bipolarnns (Cambridge University Press, Cambridge). Santoni, A., L. Terminello, E Himpsel and T. Takahashi, 1991, Appl. Phys. A 52, 299. Scalapino, D.J., 1995, Phys. Rep. 250, 329. Scalapino, D.J., 1996, Turk. J. Phys. 20, 560. Scalapino, D.J., E.Y. Loh and LE. Hirsch, 1986, Phys. Rev. B 34, 8190. Schabel, M., I. Vitomirov, G. Waddill and J. Weaver, 1991, J. Electron Spectrosc. Relat. Phenom. 56, 211.

ANGLE-RESOLVED PHOTOEMISSION STUDIES OF UNTWlNNED YBCO Schabel, M., Z.-X. Shen, J. Peng and R.L. Greene, 1995, Unpublished. Schabel, M., C.-H. Park, A. Matsuura, Z.-X. Shen, D.A. Bonn, R. Liang and WN. Hardy, 1997, Phys. Rev. B 55, 2796. Schabel, M., C.-H. Park, A. Matsuura, Z.-X. Shen, D.A. Bonn, R. Liang and W.N. Hardy, 1998a, Phys. Rev. B 57, 6090. Schabel, M., C.-H. Park, A. Matsuura, Z.-X. Shen, D.A. Bonn, R. Liang and V~N. Hardy, 1998b, Phys. Rev. B 57, 6107. Schilling, A., M. Cantoni, J.D. Guo and H.R. Ott, 1993, Nature 363, 56. Schleger, R, WN. Hardy and H. Casalta, 1994, Phys. Rev. B 49, 514. Schleger, R, R. Hadfield, H. Casalta, N.H. Andersen, H.E Poulsen, M. yon Zimmermann, J.R. Schneider, R. Liang, E Dosanjh and W.N. Hardy, 1995, Phys. Rev. Lett. 74, 1446. Schrieffer, J.R., X.-G. Wen and S.-C. Zhang, 1988, Phys. Rev. Lett. 60, 944. Schroeder, N., R. B6ttner, S. Ratz, E. Dietz, U. Gerhardt and T. Wolf, 1993, Phys. Rev. B 47, 5287. Seah, M., and W Dench, 1979, Surf. Interface Anal. 1, 1. Sharrna, R.R, T. Venkatesan, Z.H. Zhang, J.R. Liu, R. Chn and WK. Chu, 1996, Phys. Rev. Lett. 77, 4624. Shen, Z.-X., 1994, Physica B 197, 632. Shen, Z.-X., and D.S. Dessau, 1995, Phys. Rep. 253, 2. Shen, Z.-X., R Lindberg, R Soukiassian, C.B. Eom, I. Lindan, WE. Spicer and T.H. Geballe, 1989, Phys. Rev. B 39, 823. Shen, Z.-X., D.S. Dessau, B.O. Wells, D.M. King, W.E. Spicer, A.J. Arko, D.S. Marshall, L.W. Lombardo, A. Kapitulnik, R Dickinson, S. Doniach, J. DiCarlo, A.G. Loeser and C.H. Park, 1993, Phys. Rev. Lett. 70, 1553. Sheng, Z., and A. Hermann, 1988, Nature 332, 138. Shukla, A., L. Hoffmann, A. Manuel, E. Walker, B. Barbiellini and M. Pete~, 1995, Phys. Rev. B 51, 6028. Singh, D.J., and WE. Pickett, 1995, Phys. Rev, B 51, 3128. Smith, N., and E Himpsel, 1983, in: Handbook on Synchrotron Radiation, Vol. 1, ed E. Koch (NorthHolland, Amsterdam) ch. 9.

435

Soderholm, L., and G. Goodman, 1989, J. Solid State Chem. 81, 121. Soderholm, L., K. Zhang, D.G. Hinks, M.A. Beno, J.D. Jorgensen, C. Segre and I.K. Schuller, 1987, Nature 328, 604. Straub, T., R. Claessen, R Steiner, S. Htifner, V. Eyert, K. Friemelt and E. Bueher, 1997, Phys. Rev. B 55, 13473. Tjeng, L.H., C.T. Chen, J. Ghijsen, E Rudolf and E SeRe, 199t, Phys. Rev. Lett. 67, 501. Tobin, L, C. Olson, C. Gu, J. Liu, E Solal, M. Fluss, R. Howell, J. O'Brien, H. Radousky and R Sterne, 1992, Phys. Rev. B 45, 5563. Tokura, Y., and A. Fujimori, 1995, in: Spectroscopy of Mott Insulators and Correlated Metals, Proc. 17th Taniguchi Symp., Kashikojima, Japan, October 24 28, 1994 (Springer, Berlin). Tsuei, C.C., J.R. Kirtley, M. Rupp, J.Z. Sun, L.S. Yu-Jahnes, C.C. Chi, A. Gupta and M.B. Ketchen, 1995, in: Proc. Conf. on Spectroscopies in Novel Superconductors, Stanford 1995, J. Phys. Chem. Solids 56, 1787. van Harlingen, D.J., 1995, Rev. Mod. Phys. 67, 515. Varma, C.M., RB. Littlewood, S. Schmitt-Rink, E. Abrahams and A.E. Ruckenstein, 1989, Phys. Rev. Lett. 63, 1996. Veal, B., and C. Gu, 1994, J. Electron Spectrosc. Relat. Phenom. 66, 321. Virosztek, A., and J. Ruvalds, 1990, Phys. Rev. B 42, 4064. Wells, B.O., Z.-X. Shen, D.S. Dessau, WE. Spieer, D.B. Mitzi, L.W Lombardo, A. Kapitulnik and A.J. Arko, 1992, Phys. Rev. B 46, 11830. Wells, B.O., Z.-X. Shen, A. Matsuura, D.M. King, M.A. Kastner, M. Greven and R.J. Birgeneau, 1995, Phys. Rev. Lett. 74, 964. Wollman, D.A., D.J. van Harlingen, J. Giapintzakis and D.M. Ginsberg, 1995, in: Proc. Conf. on Spectroscopies in Novel Superconductors, Stanford 1995, J. Phys. Chem. Solids 56, 1797. Wu, WK., J.R. Ashburn, C.J. Torng, RH. Hor, R.L. Meng, L. Gao, Z.J. Huang, Y.Q. Wang and C.W. Chn, 1987, Phys. Rev. Lett. 58, 908. Zaanen, J., and G.A. Sawatzky, 1990, Progr. Theor. Phys. Suppl. 101, 231. Zaanen, J., G.A. Sawatzky and J.W. Allen, 1985, Phys. Rev. Lett. 55, 418.

Handbook on the Physics and Chemistry of Rare Earths Vol. 31 edited by K.A. Gschneidner, Jt:, L. Eyring and 3/LB. Maple 2001 Elsevier Science B. V.

Chapter 202 I N F R A R E D PROPERTIES OF HIGH-Tc S U P E R C O N D U C T O R S : An Experimental Overview D.N. B A S O V 1 a n d T. T I M U S K 2 l Department o f Physics, University o f California San Diego, La Jolla, CA 92093-0319, USA, 2 Department o f Physics and Astronomy, McMaster University, Hamilton, Ontario L8S 4M1, Canada

Contents

1. 2. 3.

4.

List of symbols List of abbreviations Introduction Energy scales and experimental techniques Doping antiferromagnetic insulators with charge carriers 3.1. Complex conductivity and spectral weight 3.2. Parent insulating compounds 3.3. Evolution of the spectral weight with doping 3.4. Excitations in mid-infrared frequencies Charge dynamics in the normal state 4.1. One-component versus multicomponent description of the in-plane conductivity 4.2. Models for the frequency-dependent scattering rate and effective mass 4.3. Experimental results: in-plane conductivity 4.4. Experimental results: conductivity along the Cu O chains 4.5. Experimental results: the interplane conductivity

4.6. Comparison of the optical data with transport measurements and other spectroscopies 4.7. Theoretical context 5. Electrodynamics in the superconducting state 5.1. Searching for the superconducting energy gap 5.2. Anisotropic penetration depth in cuprates 5.3. Temperature dependence of the penetration depth 5.4. Impurity effects 6. Infrared properties of HTSC in a magnetic field 6.1. The ab-plane conductivity in a magnetic field 6.2. The c-axis conductivity in a magnetic field 7. Conclusions and open questions Acknowledgements References

437 438 439 440 442 442 443 445 449 451 451 453 456 460 462

List of symbols A(g2) E E~ E~

e FL /¥t J

Bosonic spectral function Electric field Fermi energy Charge transfer gap

437

Electronic charge Lorentz force Clqtical field Current

467 470 473 473 481 486 489 493 494 496 496 498 499

438 J Jc

D.N. BASOV and 72 TIMUSK

m~ me mv Q R(0)) r t 60 T Y~ T~ Zv~w T* U G vv V

Exchange energy Josephson critical era'rent Green's function of the electronic excitation Pinning constant Carrier density Effective spectral weight In-plane spectral weight Interplane spectral weight Effective mass ab-plane effective mass c-axis effective mass Free electron mass Vortex mass Wavevector Reflectivity Complex reflectance Hopping matrix element lnterbilayer hopping Temperature Critical temperature N6el temperatnre Temperature of spin-density wave Crossover temperature Coulomb repulsion energy Condensate velocity Velocity Volume

7 A @ ,t. X, ,a,,

Damping Energy gap Phase change Penetration depth a-axis penetration depth b-axis penetration depth

G(k, 0)) k n N~nN,b Nc m*

m,,~,

,~

/~L v ~b

c-axis penetration depth London penetration depth Damping Coherence length ab-plane coherence length

P,,h Pd¢ Ps 27(0)) 2J~(co) Z2(0))

ab-plane resistivity dc resistivity Superfluid density Electronic self-energy Real part of the electronic self-energy imaginary part of the electronic self-energy

ado a(0)) a 1(m) 02(0))

dc conductivity Complex conductivity Real part of the complex conductivity Imaginary part of the complex conductivity Differential conductivity Relaxation time Vortex lifetime Magnetic flux Frequency Vortex core resonance Frequency Cut-off frequency Cyclotron frequency

a[ r vv ¢0 g2 f2o m 0)0 0)~ % COph 0)pc 0)pD 0)pk 0)p,, 0)p~

Plasma fiequency a-axis plasma frequency b-axis plasma frequency c-axis plasma frequency Drude plasma frequency Plasma frequency, Lorentzian oscillators Plasma frequency, normal component Superfluid plasma frequency

DW ESW IR KK NAFLM NMR

d-wave Extended s-wave Infrared Kramers-Kronig Nearly antiferromagnetic Fermi liquid Nuclear magnetic resonance

0)p,,

List of abbreviations ARPES ASW BCS BS CT DOS

Angular resolved photoemission spectroscopy Anisotropic s-wave Bardee~Coope~Schl-ieffer Bardeen-Stephen Charge transfer Density of states

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

439

1. Introduction

High-temperature superconductivity in copper oxides (Bednorz and Mfiller 1986) has been at the focus of solid-state physics research for more than a decade. Despite an unprecedented attack by a force that comprised a large fraction of the condensed matter community, the central issue - the mechanism of high-Tc superconductivity - still remains to be unraveled. Nevertheless, there have been several important advances in our understanding of the cuprate superconductors over the last several years. The search for the mechanism of high-To superconductivity in cuprates is in a phase dominated by experiments on wellcharacterized single crystals and thin films. This new generation of high-To materials reveals novel features which in the earlier work were obscured by defects. There is now excellent reproducibility between experiments performed on a given system by different research groups, and there is a clear tendency towards a convergence of results obtained with different experimental methods (transport, photoemission, nuclear magnetic resonance, etc). As far as the electrodynamics of cuprates is concerned, perhaps one of the most significant accomplishments is a detailed survey of the interplane c-axis conductivity at least in the YBa2Cu3Ox (Y123) and La2_ySryCuO4 (La214) series. Prior to 1992, there were several reports of the c-axis properties but no systematic study of doping and temperature dependence had been reported. The reason behind this is that single crystals of appropriate thickness have become available only recently. The interplane properties showed novel features of cuprates such as the formation of the Josephson plasma resonance at T < T~ (sect. 5.2) and of the normal-state pseudogap (sect. 5.4). The latter was found only in underdoped materials: those with T~ being reduced from the maximum value by diminishing the carrier density. While the role of the pseudogap in superconductivity remains controversial, its presence is not: there is mounting evidence coming from a variety of physical probes, all showing that the low-energy excitations in cuprates are suppressed at a temperature T* significantly exceeding T~ (sect. 4.5). Experimental results on the interplane behavior provide us with information that is essential to test several theoretical approaches to the problem of understanding high-T~ superconductivity. Single crystals of cuprates are now not only bigger in size but their quality has also improved in the recent past. A thorough study of the new generation of single crystals by a variety of transport and spectroscopic techniques has shown that the behavior of the earlier samples was often obscured by extrinsic effects. This holds true for infrared data as well. Although the ab-plane properties of YBa2Cu3Ox (x = 6-7) materials were studied systematically in 1988-1991, recent samples revealed a number of novel features in the normal- and superconducting-state electrodynamics. Infrared measurements performed on Y123 crystals prepared in different laboratories now show excellent reproducibility. Other materials such as Bi2Sr2CaCu208 (Bi2212), La2 ySryCuO4 and T12Ba2CuO6~ 6 (T1220l) also have been studied as a function of doping. The in-plane conductivity of cuprates along with the different approaches towards

440

D.N. BASOVand T. TIMUSK

the understanding of the charge dynamics in the CuO2 planes is the subject of subsects. 4.2, 5.l and 5.2. Developments in crystal preparation techniques now make it possible to grow samples with a controlled amount of impurities. Both experimental and theoretical studies of the electromagnetic response in the presence of impurities have attracted a lot of attention. The primary motivation for these activities was to test the symmetry of the order parameter in cuprates. Such tests are possible since conventional superconductors (those with an isotropic s-wave gap) and exotic superconductors (those with a strongly anisotropic order parameter) are expected to behave differently in the presence of disorder (Sigrist and Ueda 1991). We will briefly review infrared results obtained in samples with non-magnetic and magnetic impurities in sect. 5.3. It was realized quite early that the parent, undoped, compounds should be viewed as Mott insulators. More recent studies of the doping dependence have revealed the generic features between cuprates and other classes of Mott insulators. The principal trends in the evolution of the in-plane electronic conductivity with doping are in accord with the results of the calculations performed for Mott-Hubbard systems as will be described in sect. 3. Finally, very interesting results were obtained through magneto-optical studies primarily of Y123 thin films. The challenge of these experiments is that the features of interest lie close to the low-frequency boundary of infrared methods. This demanded the development of novel spectroscopic techniques employing tunable far-infrared lasers and terahertz time-domain spectroscopy. Magneto-optical measurements were used to examine the structure of the vortices, and may become helpful in resolving the controversy regarding the relaxation mechanisms in these materials. Section 6 is dedicated to magnetooptics of cuprates. There are several general review papers on the infrared properties of high-To superconductors (including those by Timusk and Tanner 1989, Tanner and Timusk 1992, and Thomas 199l), as well as a few covering more specific topics: phonons (Thomsen and Cardona 1989, Litvinchuk et al. 1994), polarons (Kastner and Birgeneau 1996), and the pseudogap state (Puchkov et al. 1996a). In this review, we will attempt to describe some recent developments in the studies of the electromagnetic response of cuprates that occurred primarily after 1992, although references to earlier work in some cases are necessary.

2o Energy scales and experimental techniques A sketch showing several important energy scales in metals and superconductors (fig. 1) may be a good starting point for a discussion of the role that infrared and optical spectroscopy have played in the study of high-To materials. Infrared and optical spectroscopy probe the elementary excitations and collective modes in the energy region from about 10 cm -1 up to 50 000 cm -1 . Both boundaries are blurred, and the actual limits depend on many factors including sample size or surface irregularities. In any event,

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

441

meV 1.0 |

I IIIIIII

10.0 I

100.0

I IIIIIII

I

I IIII1|1

t.....

I

1000.0 I

I I|lllll

I

I II

electronic transitions

phonons

I

[ .......

Josephson plasma edqe I I (pseudo)gaps in cuprates iI ....superconductorsgY gap i....... tmna, i '~'~2~'Bcs(Tc: gO K) [ scattering rate of conductin 9 carriers I I

1.0

I

I IIIIIII

I

I0.0

I IIIIIII

I

I00.0

I IIIIItl

1000.0 -I Wavenumbers, cm

I

I li

IIIII

I

I i

I0000.0

Fig. 1. Characteristic energy scales in high-~ superconductors. Infrared and optical spectroscopiesallow one to probe electronic excitationsin the frequencyrange from -20 cm-t typicallyup to at least 50000 cm 1. the accessible frequency range spans at least 3 decades in co, establishing a link between zero- or low-frequency experiments (dc transport, microwaves, tunneling) and high-energy spectroscopies (photoemission, X-rays). O f special importance for superconductors is the far-infrared part of the spectrum (co < 500cm-1), because this region includes the expected value for the superconducting energy gap 2A for materials with T~ > 10 K as well as the energy scale of the excitations (at least in some models of the phenomenon) responsible for pairing. An analysis of the spectral weight in infrared frequencies permits one to infer (through model-independent procedures) another fundamentally important parameter of a superconductor, the penetration depth )~. An interesting trend of the last few years is a broadening of the range of experimental techniques used for infrared experiments. Traditionally, infrared properties of solids have been investigated by means of Michelson interferometers (or their modifications) using black-body sources and bolometric or photoconductive/photovoltaic detectors. In these experiments reflectance or transmission is measured over a broad energy range while the optical constants of the specimen under study are obtained through KramersKronig (KK) analysis (see sect. 3.1). Michelson interferometers configured for reflectance measurements continue to be the primary workhorse in these experiments, with all system components being continuously refined (Homes et al. 1993a). An interesting novelty is the use of synchrotron light sources instead of globars or mercury lamps (Forro et al. 1990, Williams et al. 1990, Romero et al. 1992). Synchrotrons give an advantage in brightness compared to black-body sources. This brightness advantage is of special importance for studies where samples are small. This is often the case for new materials where the refinement of crystal growing techniques is not complete. Another recent innovation is concerned with ellipsometric measurements in the far infrared (Kircher et al. 1997). The principal advantage of ellipsometry is that the optical constants of a system can be obtained without K K analysis and thus measurements do not have to be extended to co ---+ 0 and co --+ ec. Ellipsometry is an extremely successful technique in the near-IR-visible-near-ultraviolet regime where high-quality

442

D.N. BASOVand T. TIMUSK

optical components are available. There are several recent reports of ellipsometric measurements in the far infrared (Herin et al. 1996, Bernhard et al. 1998-2000). These original experiments demonstrate the general feasibility of the approach and indicate the great potential of ellipsometric studies in the far infrared. At the moment these experiments require samples of significantly larger size than conventional reflectance measurements. An increasing number of experimentalists are using the technique of terahertz timedomain spectroscopy, originally proposed by Grishkovsky et al. (1990). Here infrared radiation is generated and detected using Hertz dipoles - metal stripes with a narrow gap prepared on the surface of an amorphous semiconducting substrate. A laser pulse creates photoconductive carriers in the semiconductor, thus activating the dipoles. The duration of the photocurrent is of the order of 1 ps and thus the dipole generates freely propagating radiation in the frequency region from 3 to 120 cm -1 (0.1-4 THz). By adjusting the time delay between the laser pulses which activate the source and the detector one is able to sample the time evolution of the terahertz electric field. By Fourier-transforming the detected pulse one is able to obtain both the real and the imaginary part of the conductivity of the medium through which the radiation propagates. At present this technique is used exclusively in transmission mode, and thus experiments rely on the availability of thin films. The crucial advantage of this technique is that it allows one to perform spectroscopy in the difficult frequency region below the traditional low-frequency boundary of infrared methods.

3. Doping antiferromagnetic insulators with charge carriers

3.1. Complex conductivity and spectral weight A quantitative analysis of the electromagnetic response begins with the evaluation of the optical constants of a system (complex dielectric function, complex conductivity, etc.) from the experimentally accessible quantities such as reflectance, transmission or absorption. It is convenient to discuss the response of metals and superconductors in terms of the complex conductivity o(co) = o~l(co)+ i~2(co), defined from J(co) = a(co)E(co) where J is the current vector and E is the electric field vector. All optical constants are interrelated through simple analytical expressions and contain the same physical information. The advantage of the conductivity notation is that a(co) has no singularity at co = 0. Also, an extrapolation of oh(co) to co -+ 0 yields the dc conductivity of a metal. An important characteristic of the optical constants is that they fall in a broad class of generalized susceptibilities with the fundamental property that their real and imaginary parts are connected by Kramers-Kronig (KK) integral relations (Landau and Lifshitz 1992). For the case of linear response, KK relations are model independent since they rely only on the causality principle and analytical properties of the complex susceptibilities. The physical reasoning behind the KK relations is that dissipation of energy of the

INFRARED PROPERTIES OF HTSC: EXPERIMENTALOVERVIEW

443

electromagnetic wave in a system and the phase change of the wave must be connected with each other. The KK integrals give a formal relation between the real and imaginary parts of the generalized susceptibilities. Kramers-Kronig integrals can also be used to evaluate the optical constants from the experimentally measured reflectance or transmission. Let us consider the complex reflectance index r = R 1/2 exp i°, where R(co) is the reflectivity of the solid and O(co) is the phase change between incident and reflected waves. KK integrals provide the connection between O(co) and R(co): 2co f ~ lnr(f2) - In r(co) O(co) = ~ - J0 ~ _--~ do).

(1)

A combination of R measured experimentally and O expressed in terms of R with eq. (1) allows one to obtain any set of optical constants (see Born and Wolf 1980). Analogously, the phase change can be inferred from transmission. An important consequence of the K K relations is the possibility of formulating numerous sum rules for the optical constants (Smith 1985). In particular, the oscillator strength sum rule is essential for the quantitative analysis of the optical data: fo ° (71(co) dco _ co~ _ 4 ~ n e 2 8 m* '

(2)

where n is the carrier density and m is the carrier mass. If integration is performed up to to some cut-off frequency coo, eq. (2) yields the effective mass of carriers m* [see also the discussion of eq. (4) in subsect. 3.3]. This sum rule is particularly important for superconductors because it can be used to evaluate the density of superconducting condensate as will be explained in sect. 5. A sum rule for the imaginary part of the conductivity establishes a connection between the spectrum of o2(0)) and the dc conductivity of a solid which could be measured independently using a resistance probe: f0 °~ (x2(co)dco = ~ ~Oac -

(3)

The practical use of eq. (3) is to test the impact of extrapolations of the experimental reflectance to co ---+0 and co ~ oc. Such extrapolations are always required in accordance with eq. (1). However, if the energy region of the experimental data is sufficiently broad, extrapolations (in particular to co - + ec) create only negligible error, and Oac obtained from eq. (3) should be close to the directly measured value. 3.2. Parent insulating compounds" It is well established that superconductivity in cuprates occurs when the CuO2 planes, which are shared by all high-To compounds, are doped away from half filling either with

444

D.N. BASOVand T. TIMUSK

holes or with electrons. The parent undoped compounds are antiferromagnetic insulators with TN ranging from 400 to 1000K in different series. The dominant feature in the in-plane (E I] ab) response of undoped cuprates is a sharp onset at co = 1-2 eV in al (~) which originates from a charge transfer (CT) gap (Fnjimori et al. 1987, Tokura et al. 1990). Differences in the magnitude of the CT gap were ascribed primarily to different coordination of oxygen sites surrounding a Cu site. With increasing number of apical oxygen sites the magnitude of the gap is enhanced. The absorption features at 1-2eV are prominent for light polarized along the C u O 2 planes, while the c-axis response (E II c) shows no sharp structure in the CT region. This anisotropic character of the charge-transfer excitations suggests that they are mainly associated with transitions fi'om O 2p to Cu 3dx2_y2 states (Fujimori et al. 1987, Tokura et al. 1990). In all compounds the absorption edge appears to be broadened. Detailed investigations of the temperature dependence of the 2 eV threshold in LazCuO4 by Falck et al. (1992), accompanied by photoconductivity experiments in the same energy range, suggest that the broadening could be attributed to coupling of the CT excitations to optical phonons. The observation of the CT gap should be contrasted with the predictions of band theory. Local-density-approximation (LDA) calculations performed for a number of undoped materials predict these systems to be metals since the Cu 3d and O 2p orbitals form a conduction band which is half-filled (for a review see Pickett 1989). The spin-polarized version of this band theory is not sufficiently accurate to yield an antiferromagnetic state of the insulating compound (Pickett et al. 1992). Thus LDA calculations fail to account for the two principal features of undoped materials: the insulating gap and antiferromagnetic ordering. However, despite these serious problems these calculations do yield accurate values for the Fermi surface crossings as observed by angle-resolved photoemission (for a review see Pickett et al. 1992; see also ch. 201 of this Handbook). The Hubbard model, on the other hand, successfully describes the antiferromagnetic insulator state. The Coulomb repulsion between Cu ions leads to a splitting of the 3d band into lower and upper Hubbard bands with an energy U separating the bands. This energy U is the cost of creating a doubly occupied Cu site in the upper Hubbard band. Another important control parameter of the Hubbard model is the inter-site hopping matrix element t. If the spins on neighboring sites are oppositely oriented, they can hop virtually from one site to another, lowering the energy of the system by a term proportional to -t2/U but only if (as required by the Pauli principle) their spins are antiparalM. This is equivalent to an exchange interaction with J = 4tZ/u, and gives rise to antiferromagnetism. In the cuprates the oxygen 2p band is located inside the Hubbard gap, which leads to a situation where the CT gap, Eg, is less than U. Adopting the classification proposed by Zaanen et al. (1985) this case is called a "charge-transfer" instflator. An appealing feature of the Mott-Hubbard scenario is that it is capable of yielding a variety of ground states including paramagnetic (metallic and insulating) states as well as antiferromagnetic (also metallic and insulating) ones (for a review see Georges et al. 1996). The evolution of the CT gap and the evolution of the spectral weight with

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

445

doping, which will be described in the next subsection, strongly support the soundness of the Mott-Hubbard approach with respect to the optical properties of cuprates. 3.3. Euolution of the spectral weight with doping In the L a 2 C u O 4 system the substitution of La 3+ with Sr2+ creates holes in the CuO2 planes. Metallic or, more precisely, conducting behavior in the CuO2 planes was observed in La2_yS13,CuO4 materials for dopings as low as y = 0.04. In Nd2 yCeyCuO4, Nd 3+ ions are substituted with C e 4+, which leads to electron-like doping of the CuO2 planes. A somewhat different type of doping is realized in the YBa2Cu3Ox system where excess oxygen ions fill the one-dimensional Cu-O chains. In any event, in all of the cuprates the CuO2 planes remain chemically unchanged in the process of doping: the dopants alter the composition of the so-called charge-reservoir layers [La-O, Nd-O, or Cu-O chains in Y123 and YBa2Cu408 (Y124)]. The conducting properties of the cuprates are primarily associated with the CuO2 planes (see sect. 3.2). Therefore, disorder in charge-reservoir layers, which is inevitable in metallic phases, does not necessarily introduce extra scattering in the ab-plane transport. The direct involvement of the charge reservoirs in transport and conductivity phenomena is not well studied. The Y123 and Y124 materials form an exception; there the contribution of the reservoir-layer Cu-O chains can be studied in detwinned crystals. The chain dc and ac conductivities in the normal state and the superfluid density at T < T~ have been established by several different groups of experiments (Basov et al. 1995a, Tallon et al. 1995, Edwards et al. 1995, Sun et al. 1995). A typical variation of the in-plane optical conductivity with doping is shown in fig. 2 for the La2 y S r y C u O 4 system Llchida et al. 199l). The principal result is that f o r y > 0 the

o8!

1,5

La2_xSrxCuO4

La2_xSrxCuO4

_0.34

E 7 1.0

3"

X=O

v

_0.15

~

C, 3

g o.5

/

0.6

_0.20

0.10

0,02-

~\

///

~ f ~ o . ~ s .. ......

z

0.20 / / Z , / ' * ' 0,t'5

-" ~"

0.~

.//~/--o.o2

O.O2

O0

1

2h~ (eV)3

4

O0

t

2

h.)

3

4 (ev)

Fig. 2. The real part of the ab-plane conductivity (left panel) and the effective spectral weight determined from the conductivity using eq. (4) for La214 crystals, after Uchida et al. (1991).

446

D.N. BASOVand T. TIMUSK

energy region below the CT gap is filled with states at the expense of the spectral weight associated with the CT excitation in the undoped counterpart. Notably the conductivity at 1.5 eV is almost unchanged by doping. This isosbestic behavior at co ~ 1.5 eV is clearly observed in Nd2_yCeyCuO4 materials (Uchida et al. 1991, Tanner et al. 1996). Another characteristic of the doped samples is that a feature due to the CT gap persists in the spectra up to quite high levels of doping when the system is well in the metallic regime. To quantify the changes in the optical conductivity induced by doping it is instructive to explore the behavior of the effective electron number Neff(co): Neff(co)

=

2me V :re 2

'

o'I (g2) dr2,

(4)

where me is the free electron mass and V is the volume of the unit cell. Neff is proportional to the number of electrons participating in the optical excitations with energies less than co. The right-hand panel of fig. 2 shows the frequency dependence of Neff(co) for the family of La2_ySryCuO4crystals. In undoped samples the major part of the spectral weight is accumulated from energies exceeding Eg, whereas in the region co < 1.5 eV, Neff(co) develops a plateau. With increasing y, the low-frequency spectral weight builds up rapidly. For y < 0.1, Neg(co < 1.5 eV) varies nearly linearly with y. This growth of Neff occurs at a faster rate than one would expect within the assumption that each Sr ion donates 1 hole per unit cell with an effective mass equal to the free-electron mass me. As shown by S.L. Cooper et al. (1990), a faster than expected growth of Neff is also found for electron-doped Pr2-yCeyCuO4 materials. The slope of the Neff versus y dependence is even steeper for electron-doped Pr2 yCeyCuO4 and Nd2 yCeyCuO4(S.L. Cooper et al. 1990, Arima et al. 1993) than for hole-doped La214. It follows from fig. 2 that doping affects electronic excitations located below 3 eV. Indeed, Ne~'(co) spectra converge at about 3 eV for all y < 0.2. An unexpected result of fig. 2 is that at y > 0.2, the total spectral weight decreases so that Neff shows a nonmonotonic dependence as a function of y, with the maximum located around y = 0.2 in the La214 series. Puchkov et al. (1996b) have analyzed the variation of the spectral weight with doping in a variety of high-T~ materials. Their conclusion is that Neff either saturates or even decreases above optimal doping which was defined as the carrier density that corresponds to the maximum critical temperature. Thus, the evolution of the in-plane spectral weight in cuprates with hole/electron doping reveals the following principal trends: (i) the CT gap is filled with states which were located above Eg in undoped materials; (ii) the total spectral weight (a(co) integrated up to 3 eV) is unchanged at small or moderate dopings and is suppressed in the overdoped regime (y > 0.2); (iii) the low-frequency Neff (a(co) integrated up to 1.5 eV) increases faster than y in underdoped samples and is suppressed in the overdoped regime. Points (i) (iii) should first be contrasted with the behaviour of a Fermi liquid with a large Fermi surface. Doping with holes reduces the size of the Fermi surface and one

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

447

0.20 /x

= 0.25

/ x = 0.375 0.15

Hubbard U/t=lO

0.10 b

x=O

0.05

0.00 0

5

10

o)/t

15

Fig. 3. The optical conductivity calculated within a one-band Hubbard model on a 4x4 cluster (Dagotto 1994). Parameter x stands for the hole-doping fraction. Calculations reproduce the principal trends of the experimental data on the evolution of the electromagnetic response with doping. In particular, theoretical spectra reveal: (i) the transfer of the spectral weight from the CT excitation to the low-frequency region; (ii) development of "mid-infrared" absorption at light doping (x = 0.125) and (iii) growth of the Drude-like feature at higher dopings (x > 0.2). Note that the isobestic point at oct = 5 is also reproduced.

expects to see a decrease o f spectral weight with doping x as 1 - x rather than the observed dependence that is proportional to x. On the other hand, these features seem to be generic for a broad variety of Mort insulators, including cuprates (S.L. Cooper et al. 1990, Orenstein et al. 1990, Arima et al. 1993), nickelates such as La2 ySryNiO4 (Katsufuji et al. 1996), transition-metal oxides such as (La,Pr, Nd, Sm,Y)2 ~CaxTiO3 (Katsufuji et al. 1995), spin-ladder compounds (Osafune et al. 1997) and vanadium oxides (Thomas et al. 1994, Rozenberg et al. 1995). The Mott Hubbard picture, which accounts for the AF insulator state o f undoped compounds, also helps us in understanding the changes in the redistribution o f the spectral weight with doping. The Hubbard model has two principal energy scales, U and t, and both can give rise to observable features of the optical conductivity. In the half-filled case, one expects to find an absorption band centered at U. Simple charge-counting arguments suggest that the spectral weight associated with this interband transition will decrease with hole doping while its energy remains unchanged (Stephan and Horsch 1990). This is clearly shown in fig. 3 where results o f calculations for a one-band Hubbard model are presented (Dagotto 1994). Dagotto obtained spectra o f c~l(co) that reproduce the principal trends o f the experimental data for cuprates. The spectral weight removed from the CT transition is recovered at lower energies on the order o f t. Note that the conductivity is identical for the hole doping fractions x = 0.25 and x = 0.375. This corresponds to a non-monotonic evolution of the spectral weight in the intragap region. This latter result is reproduced in several other calculations starting from a Mott-Hubbard Hamiltonian (Stephan and Horsch 1990, Jarell et al. 1995, Kajueter et al. 1996). Another noticeable feature o f the theoretical spectra presented in fig. 3 is an isobestic point at colt = 5. If t is taken to be equal to 0.3-0.4 eV as some calculations suggest (Hybertsen et al. 1990, Bacci et al. 1991), the calculated position o f this point and of the CT gap Eg ~ 6t would

448

D.N. BASOV and T. TIMUSK

2000

-~ I: A I/ , [-!i\ ]i

~E 0

,/..-o.,o

& ~oI- )I ~i~/K / / / /

i

v

•~ 100C x=0.30

/

H,'I

l h1

Ii;it

11i l

~,(,,o

/~//

,/7

o.2o

//o.,t o.lo

u0 "N

//:1

//~:;:8:1~ l

/

.,,y/,

j~"

/

1

/

o

0 Photon Energy (eV)

O.OOS

'

i

==°'~°~I~-~ •

i

~

i

I'~2-xSrxCu04 0.004 ~: 0.003

0.25

0.002

0.001 0

0.2 ~o (~v)

0.4

Fig. 4. The c-axis optical conductivity (top panel) and the effective spectral weight determined from eq. (4) for a series of La x xSrxCuO4 single crystals (Uchida et al. 1996).

agree with experiment. Finally, by taking into account hybridization effects it was possible to reproduce the asymmetry between the doping dependence of the spectral weight for holes and for electrons (Eskes et al. 1991, Meinders et al. 1993, Tanner et al. 1996). The doping dependence of the interplane c-axis conductivity has been studied so far only for La214 (Uchida et al. 1996) and YBa2Cu3Ox (Homes et al. 1995a,b, Tajima et al. 1997). A brief examination of fig. 4 (the c-axis spectra for La2_ySryCuO4) reveals an important difference between the in-plane and interplane properties. First, the CT gap at 1.5 eV is not seen in the interplane conductivity of La2CuO4. The electronic contribution to ~c(~o) is strongly suppressed at :o < 5 eV The anisotropy of Neff integrated up to 2 eV is as high as 10.

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

449

Upon doping, a certain part of the spectral weight in the 5-6 eV region is transferred to low frequencies. The c-axis spectral weight No(y) integrated up to 2 eV increases linearly up to y = 0.2 and then saturates at higher dopings. This tendency resembles the behavior of N,b(y) (fig. 2). However a closer comparison of Nob(y) and N,:(y) shows that the relative fraction of the spectral weight transferred to low frequencies is larger for the interplane conductivity. As a result, the anisotropy of Noh/Nc decreases from 10 in undoped compounds down to 3 in overdoped materials. The anisotropy of the spectral weight should be distinguished from the anisotropy of the free-carrier (Drude) plasma frequencies C02D = 4arne2/m*, where n is the concentration of free carriers and m* is their effective mass. That is because the integral in the definition of Neff(co) includes the contributions of both free and bound carriers which both give rise to the conductivity at co < Eg. The analysis of the possible mechanisms behind the absorption in the intragap region is the subject of the next subsection. We conclude this subsection by noting that Nd2CuO4_z, La2CuO4 ~z and some other materials could be doped not only by substitution of Nd or La with Ce or Sr but also by partial removal or addition of oxygen. Quijada et al. (1995) have studied the conductivity of oxygen-doped La2CuO4.12 an Sr-free material with Tc as high as 40 K. Their results for the in-plane conductivity are in agreement with the data of Uchida et al. (1991) for the Sr-doped crystal near the optimal doping. This suggests that there is an equivalence between different approaches that introduce carriers in the CuO2 planes. 3.4. Excitations in mid-infrared frequencies The detailed analysis of the optical conductivity in the CT gap region reveals several distinct features which are particularly well resolved in samples with dilute concentration of charge carriers (figs. 2 and 5): (i) a broad resonance at 0.5 eV, which shows considerable softening as doping increases; (ii) a resonance at 0.25 eV; (iii) a Drude-like peak at co = 0. The existence of well-separated absorption features in the frequency dependence of crab(a)) is the basis for multicomponent analysis of the conductivity. Within this approach (Tanner and Timusk 1992) the conductivity is assumed to have the form

2

(O2k

%DT + l, o(cok- 2i or O.

a(~o) - 4 ~ ( 1 + i o r )

(5)

In eq. (5) the first term stands for the Drude free-carrier conductivity with plasma frequency COpD and scattering rate l/r. The second term accounts for the additional contributions to the conductivity that are modeled with Lorentzian oscillators centered at frequency ~oi,with plasma frequency ~Opl,and damping Y01,-These additional contributions are assigned to two-magnon excitations, interband transitions, polarons and impurity states. Thomas et al. (1992) have shown that peaks at 0.25 eV and 0.5 eV are found in a variety of cuprates including YBa2Cu306+x, La2CuO4+y and Nd2CuO4_y. They analyzed the positions of these peaks and compared them with other related energy scales in these

450

D.N. BASOVand T. TIMUSK I

250

I

Ej

I

E! l

I

Ba2YCu306+y

I,ld2C,~O,_y/ I

2OO

'e 0

L

0.25

0.5

La2CuO4+y

0.75

1.0

1.25

E (eV)

Fig. 5. Excitations in the mid-infrared range in lightly doped crystals of Ba2YCu3Q+~,, Nd2CuO4 v and La2CuO4~!,(Thomaset al. 1992). In lightly doped crystals one can easily distinguish between severaldifferent absorption features in the mid-infrared.

materials. The frequency positions of the peaks do not scale in a simple way with the magnitude of the CT gap Eg. The location of the 0.25 eV feature is close to the Cu-Cu exchange energy J Interpretation of the structure in mid-infrared in terms of polarons has attracted a lot of attention. As pointed out by Falck et al. (1992), a polaron absorption model quantitatively describes the broadening of the charge-transfer excitation peak. Bi and Eklund (1993) applied the polaronic model to the analysis of the conductivity for La2 ySryCuO4 + ~ and for La2_ySryNiO4+6. Polaron-transport theory provides reasonably accurate fits to the measurements of the photo-induced infrared conductivity in T12Ba2Ca0.98Gd002Cu2Os, YBa2Cu306.3 and LazCuO4 (Mihailovic et al. 1990). The mid-infrared conductivity in undoped and lightly doped cuprates has also been discussed in terms of polarons by Dewing and Salje (1992), by Alexandrov et al. (1993) and by Yagil and Salje (1996). Calvani et al. (1996) have argued that the polaronic features persist in superconducting samples of Nd2 .yCeyO4 a- Dolgov et al. (1997) have fitted the ab-plane conductivity data (Gao et al. 1993a) for superconducting La214 using a broad Lorentzian term that they associate with a polaron band. These and many other experiments suggest that electron-phonon interaction is an important factor in the electromagnetic response of cuprates. This interaction may be responsible for peculiar features in the conductivity spectra such as anomalous coupling of the in-plane charge carriers to the c-axis longitudinal modes first observed by Reedyk and Timusk (1992) or

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

451

a broad peak at 340 cm -1 seen in the c-axis spectra of underdoped Y123 crystals (Homes et al. 1993b, 1995b, Basov et al. 1994a, Hauffet al. 1996). However, there is very little evidence to support the idea that interaction with the lattice modes in any form is the dominant mechanism of superconductivity at high critical temperatures. Allen (1992) and Emery and Kivelson (1995a) have pointed out a paradoxical situation with the electron-phonon interaction in cuprates and other doped insulators. In models based on the polaronic interpretation of the optical conductivity, or in models explaining the temperature dependence of the phonon frequencies (for a review see Litvinchuk et al. 1994), the electron-phonon interaction would be quite strong. However, there are no signatures of strong electron-phonon coupling in the transport properties. The in-plane resistivity of many cuprates varies linearly with T over the temperature region from Tc up to 1000 K with a zero intercept on the temperature axis, which is inconsistent with scattering of charge carriers by phonons where an intercept on the temperature axis of the order of 0D is expected. Perkins et al. (1993) have observed a feature at 0.35 eV in the absorption of the undoped materials La2euO4, Sr2CuO2C12, Nd2CuO4 and Pr2CuO4. Recently this weak absorption was also found in YBa2Cu306 (Grtininger et al. 1996, Tanner et al. 1996). According to Lorenzana and Sawatzky (1995) this peak originates from the creation of a two-magnon quasibound state which is coupled to an optical phonon. Magnons are not expected to give rise to infrared absorption in materials with inversion symmetry such as La2CuO4 and other systems studied by Perkins et al. (1993). However, in a process involving a phonon and two magnons the symmetry of the lattice is effectively lower, and this absorption becomes allowed. Calculations reproduce the lineshape of the main band and indicate that the weak structure in the region of 0.6-1.2 eV may originate from higher-order processes. The magnitude of the superexchange energy J = 0.121 eV used in these calculations is close to estimates based on Raman scattering (Lyons et al. 1988). Also, the theory predicts a very different lineshape for structurally similar La2NiO4 in which Ni (spin 1) replaces copper (spin 1). The experimental data obtained by Perkins et al. (1995) for Ni-based compounds indeed reveal different lineshape for the mid-infrared absorption.

4. Charge dynamics in the normal state

4.1. One-component versus multicomponent description of the in-plane conductivity As the doping level increases, the sharp features in the mid-infrared energies acquire an increasing amount of oscillator strength (fig. 2). It is also more difficult to resolve these features in the superconducting phases as they blend into the broad Drude peak centered at zero frequency and a featureless background extending throughout the whole mid-infrared range. The La214 system is perhaps the only compound where well-structured midinfrared absorption is observed in superconducting phases (Uchida et al. 1991, Startseva et al. 1999a,b). In most other cuprates, including Y123, Y124, Bi2212, Nd214, T12201 and others, the conductivity spectra obtained for metallic and superconducting phases

452

D.N. BASOVand T. TIMUSK

show smooth mid-infrared conductivity without any characteristic absorption bands. At frequencies below 500 800 cm 1 this behavior evolves into a Drude-like feature. It is important to emphasize that the qualifier "Drude-like" is used here only to indicate that the conductivity shows a peak centered at zero frequency which is typical for the response of a Drude metal. The detailed analysis of the lineshape of this feature shows that it does not have the Lorentzian shape suggested by a simple Drude formula (first term in eq. 5). Indeed, the fall-off of C~l(co) in Y123 and all other cuprates is significantly slower than is prescribed by eq. (5). As discussed in detail by Orenstein et al. (1990), it is impossible to fit the conductivity with a simple Drude formula. If the scattering rate in the Drude equation is set at a low value to reproduce the shape of the low-frequency peak in Crl(co), then the fit is completely wrong at higher frequencies. I f the width of the Drude peak is chosen to be anomalously broad in accord with the behavior of the conductivity at co > 600 cm -~, then the model yields the wrong magnitude of the low-co behavior and reveals strong disagreement with the dc conductivity. The smooth mid-infrared continuum that follows the peak at co = 0 is found not only in doped cuprates, but also in ruthenates, cobaltates, and nickelates (Bozovic et al. 1994, Gervais et al. 1988), and thus may be regarded as another generic feature of the doped Mott insulator. Two principal approaches are commonly used to describe the deviations of the conductivity in mid-infrared frequencies from the Drude formula: a multicomponent and a one-component model. As pointed out in the previous subsection, the basic argument in favor of the multicomponent scenario (eq. 5) is that the spectra obtained for lightly doped materials reveal distinct and well-separated absorption resonances which evolve into a smooth background that blends in with the Drude band as one proceeds with doping. On the contrary, within the single-component approach it is assumed that the sole cause of the frequency dependence of a(co) is the response of itinerant carriers which acquire a frequency-dependent scattering rate 1/r(co) and a frequency-dependent mass m*(co) as the result of strong interactions in a system. This formalism is commonly referred to as the "extended Drude model": a(co, T) -

1 CO2D 2st 1/7:(co, T) - ico[m*(co, T)/m]"

(6)

Both single- and multicomponent approaches have their advantages and important shortcomings. The former picture offers a better treatment of the response at low frequencies. But its applicability at energies of the order of 1 eV is meaningless since at these high energies there are additional (interband) contributions to a(co). The effective mass m*(co) becomes negative at frequencies above 7000cm ~, clearly supporting the view that the high-energy part of the spectrum is influenced by additional excitations (Puchkov et al. 1996a). Band-structure calculation predict a rich variety of interband transitions in the frequency range starting from 1000cm -1 (Maksimov et al. 1989). However, it is not easy to separate contributions to the conductivity arising from "free" and "bound" carriers because there are no experimentally detectable signatures of the crossover from one behavior to another.

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

453

In an attempt to come up with a more rigorous procedure to separate the "free" carrier response from mid-infrared term, Romero et al. (1992) have analyzed the superconducting state response of Bi2Sr2CaCu208 single crystals. They assumed that at T << T~, all free carriers condense and their spectrum consists simply of a delta function at co = 0. This is a reasonable picture since various experiments suggest that the in-plane response of high-To superconductors is in the clean limit (Bonn et al. 1992). The condition for the clean limit is that the residual elastic scattering rate due to impurities is much smaller than the magnitude of the superconducting gap. Thus, the remaining contribution to or(co) may be assigned to the mid-infrared term. Following this logic, the difference between the spectra in the normal state and in the superconducting state, o'((co, T) = [~l(co, T > T~) crl(co,T << T~)], yields the free-carrier contribution to the normal state conductivity. Several qualitative arguments suggest that it is probably safe to employ the singlecomponent picture to analyse the data at energies below 4000 cm -~. First, as noted by Thomas et al. (1988), the number of carriers that one obtains using the sum-rule analysis for ol(co) is consistent with estimates from chemical-valence arguments for the carrier density, provided the integration in eq. (2) is carried out up to a cut-off frequency ~8000 cm -1. This is in accord with the notion that the response of conducting carriers is predominant at frequencies at least up to 4000 cm -1 . A somewhat similar estimate of the high-frequency limit of applicability of the one-component picture can be deduced indirectly from the fact that the temperature dependence of ~i(co) is also restricted to frequencies below 3000-4000 cm -1. It is natural to assign the temperature-dependent part to "free" carriers while the temperature-independent contribution may be associated with mid-infrared terms due to bound carriers. Recently, Atkinson and Carbotte (1997) calculated the interband contribution from chain-plane transitions and found that in the case of Y123 the contribution to the c-axis response was substantial, but that for the ab-plane conductivity was a minor factor at co < 2000 cm -1, suggesting that the main response comes from the intraband (Drude) processes. 4.2. Models for the frequency-dependent scattering rate and effective mass In using the one-component model one should separate two steps of this analysis: (i) the extraction of the frequency-dependent scattering rate 1/r(co, T) and of the mass enhancement m*/me from the experimental data; and (ii) the assignment of particular physical meaning to the spectra of 1/T(CO, T) and m*/me. The former step is a legitimate procedure since 1/z'(co, T) and m*/mo can be viewed as the real and imaginary parts of another response function, the memory function M(co) (Allen and Mikkelsen 1976, Mori 1965, G6tze and W61fle 1972). Expressing Re (M(co)) and Im(M(co)) through the complex conductivity one obtains the following relations: 1

T(CO)

_ ~gpD~

4 ~ Ke

1 ,

(7)

454

D.N. BASOV and T. TIMUSK m*

~0p2D 1 1 m0) (o(~)4~

"

(8)

me

Since o(co) is causal, the scattering rate and the mass enhancement are not independent, and are connected through the Kramers-Kronig integral equation:

m e

X

Jo

~~'~2_ to d£2

1.

(9)

From the discussion above it follows that 1/T(co) and m*/me c a n be viewed as optical constants. For example, 1/v(co), if multiplied by the constant plasma frequency, can be viewed as a kind of optical resistivity (reducing to the dc resistivity at zero frequency). The spectra of 1/r(w) and re*line describe the response of a system to the electromagnetic field with the same success as any other pair of optical constants such as the conductivity or the dielectric function. Clearly, 1/r((o) and re*line do not contain new physical information as compared to the complex conductivity, but this description turns out to be much more enlightening in several cases as will be shown below. However, the interpretation of 1/r(a)) and m*/me obtained from the one-component analysis is more difficult. Obviously, it is tempting to assign the frequency-dependent effective mass and the scattering rate to the real and imaginary parts of the electronic self-energy 2;(co) = 2Jj (co) + iZ2(co) entering the equation for the spectral function of the electronic excitation A(k, (o):

A(k, (o) =

1 ii m

G(k,

co)l

:

1 ImZ(m) Jr (co - ci, - Re ~v((D))2 4" (Im ~'((O)) 2"

(10)

Here, G(k, co) is the Green's function of the electronic excitation. A connection between the effective mass and the scattering rate with the self-energy, Z((o) = 1

m*(~o)

i

m

r(co)'

(11)

can be justified for a Fermi liquid in the limit of zero temperature. However, even in this simple case one should distinguish transport mass or relaxation rate, which are calculated theoretically, from the optical mass or relaxation rate which can be obtained from infrared experiments (Allen 1971, Shulga et al. 1991). A more complex situation arises in cuprate high-To superconductors where applicability of the Fermi liquid picture itself is questionable (Anderson 1987, Littlewood and Varma 1992, Emery and Kivelson 1995a). Keeping in mind possible caveats with respect to the interpretation of the results inferred from the extended Drude model, we find that this approach to the data analysis is extremely fruitful and provides important insights into the charge dynamics in cuprates. It is instructive to examine the behavior of the frequency-dependent scattering rate and of the effective mass within simple models. Here we restrict ourselves to one model developed

INFRAREDPROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

455

by Allen (1971) for electron-boson scattering. For the case of finite temperatures, this model was generalized by Shulga et al. (1991): ~(co, T) = ~ f 0 °~ dg2A(g2) I2cocoth ( ~ T ) -(co+g2)coth ( c o + g2) 1 1

+ (CO- g2) coth ~ -

(12)

+ rimp

In this equation A(g2) is a phonon (or more generally bosonic) density of states weighed by the amplitude for the large-angle scattering on the Fermi surface. Figure 6 presents the results of calculations using eq. (12) (Puchkov et al. 1996a). The form of bosonic spectrum used for the calculations is a rectangle confined between 0)o/2 and 20o/2 (top panel). The lower panels show the real part of the conductivity, the relaxation rate and the mass enhancement. An obvious result of fig. 6 is that the effective mass at low temperatures is strongly enhanced by the electron-boson interaction. "Heavy" carriers dragging a bosonic cloud are more difficult to scatter, which may explain the suppression of 1/t(co) and of the width of the Drude-like feature in al(co) at low temperatures. This situation is known from studies of heavy-fermion materials where anomalously high effective masses lead to the formation of an extremely narrow mode in the spectra of cq(co -+ 0) (Bonn et al. 1988, Sulewski et al. 1988, Awasthi et al. 1993). Another way to understand the suppression of 1/T(co) is to turn to Alien's formula for the scattering rate in the limit of T = 0 when 1/r(co) ~ f o dg2A(g2). At low frequencies (co < 0O/2), the scattering rate 1/t(co) is small since there are no energetically accessible scattering excitations in this frequency range. As soon as the scattering channels become available (co > COo),the scattering rate rapidly increases. However, by increasing the frequency above the high-energy cut-off (30)o/2) one does not gain any additional scattering channels, which accounts for the saturation of 2/r(~o) in the high-energy part of the spectra. Recently, Carbotte et al. (1999) inverted eq. (12) in an attempt to extract the frequency dependence of the spectral function A(g2). They found that A(g2) peaks at approximately 40meV in the optimally doped Y123 compounds. This energy coincides with the frequency or the resonant mode at (Jr, Jr) found in neutron scattering experiments. Further analysis revealed common trends in temperature and doping dependence of the neutron mode and the spectral function A(g2). These results were interpreted in terms of strong coupling between the charge carriers and magnetic degrees of freedom. It is important to mention one serious deficiency of the extended Drude analysis. This formalism is based on an isotropic version of Fermi-liquid theory which is highly questionable in cuprates where the electronic structure is in fact very anisotropic in the ab-plane. One approach that addresses this problem is to first calculate the conductivity, using a proper anisotropic theory with k-dependent Fermi velocity and scattering rate l/T, and then use the real and imaginary parts of the calculated conductivity to find the effective scattering rates 1/T(co) and rn*. Some steps in that direction have been made by Stojkovic and Pines (1996) and Branch and Carbotte (1998).

l[j

456

D.N. BASOV and T. TIMUSK

o

T

T

T

q

}

{

{

+

~l

=

1

......

o.1

----"=-.-..- .

o.ol 2

.

.

.

.

.

.

.

.

.

.

.

.

.

4-

t .

.

.

.

.

.

.

.

~

. . . .

o ~ . . ~

..............

.

.

1 =001

.

i

.

+ . . . . .

~

. . . . . . . .

~

. . . .

, .

.

.

.

.

.

.

.--.-.-.--.-.-.--.2.-_

0

.......

T=COo T=0.5O~

4 3

~

~E

..... 2

1

.

.

.

.

...........

...._ -'".~":~

.......

T=0.2503o

........

T=0.1250)o

.

~". . . . . . . . . .

~. . . . . . . . . . .

1

2

i" 3

..................... 4

Fig. 6. Electron-boson model calculations (from Puchkov et al. 1996a). The top panel shows the bosonic spectral density, here a "square" spectrum; the next panel shows the optical conductivity; the third panel shows the scattering rate, and the bottom panel shows the mass renormalization. The coupling constant is equal to 1. 4.3.

Experimental results." in-plane conductivity

The one-component analysis o f the optical conductivity o f YBa2Cu3Ox was first applied by Thomas et al. (1988) and later employed by many groups to describe the response o f a variety ofhigh-Tc superconductors (Collins et al. 1989, Orenstein et al. 1990, Schlesinger et al. 1990, Rotter et al. 1991, Thomas et al. 1991, Bucher et al. 1992, S.L. Cooper et al. 1992, 1993a, Wachter et al. 1994, E1 A z r a k et al. 1994, Basov et al. 1996, Puchkov et al. 1996c). The principal results o f this analysis, which have attracted a great deal o f attention can be formulated as follows: (i) the spectra o f 1/T(co) taken at room temperature are nearly linear with co over a broad range o f frequencies and do not show any indications

1NFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

457

of saturation at frequencies as high as 0.5 eV; (ii) the absolute value of 1/T(co) is on the order of the frequency. In particular (ii) challenges the applicability of the Fermi-liquid theory to high-Tc superconductors, since one of the basic assumptions of this theory is that the quasiparticle damping is much lower than the quasiparticle energy. Experimental results clearly suggest that this assumption is violated in (super)conducting phases of cuprates. From (i) and (ii) it appears that charge carriers in cuprates are heavily dressed due to strong interactions which are likely to be of electronic or magnetic origin. A detailed discussion of the charge dynamics in cuprates from underdoped to overdoped regimes was reported by Puchkov et al. (1996a). Since it is difficult to access all doping regimes using one family of cuprates, they used Y123, Y124, Bi2212 and T12201 materials which in combination allow one to follow the general trends in the evolution of the electromagnetic response throughout the entire phase diagram. Below we review the main results of their studies. Underdoped cuprates. The electromagnetic response of these materials has an additional temperature scale or crossover temperature T* which can be as high as 300K in strongly underdoped materials but which is diminished as doping increases toward optimal. The scattering rate is nearly linear with co at T > T* (fig. 7), but at T < T*, 1/T(co) is suppressed at low frequencies. This suppression corresponds to the rapid narrowing of the Drude-like feature in the conductivity spectra. A decrease of the scattering rate occurs only at low frequencies (co < 700 cm -1) while at higher energies 1/T(co) is nearly temperature independent. The remarkable feature ofunderdoped samples is that electromagnetic response in the pseudogap state which is realized at Tc < T < T* is very similar to that in the superconducting state. The bottom panel of fig. 7 shows that interactions in underdoped materials lead to an enhancement of the effective mass of conducting carriers by as much as a factor of 5. Optimally doped samples. Close to optimal doping, the threshold structure in the spectra of 1/T(co) at T > Tc is either strongly suppressed (as in Bi2212 and T12201) or disappears completely (Y123) (fig. 8). The normal-state scattering rate in these crystals is a linear function of frequency and of temperature (Gurvitch and Fiory 1987). However, the behavior of 1/T(co) and m* in the superconducting state is very similar to the behavior in the pseudogap state for underdoped crystals. The diminishing of 1/v(co --+ 0) in the superconducting state is consistent with the observation of an abrupt drop of the scattering rate in the optimally doped samples at T < Tc (Bonn et al. 1992, Romero et al. 1992). One difference from underdoped materials is that 1/r(co) acquires a weak temperature dependence at high frequencies. The effective mass of conducting carriers is somewhat reduced compared to the underdoped phases and does not exceed a factor of 3 in the normal state. Overdoped crystals. In overdoped samples, a threshold structure in the 1/r(co) spectra is not observed (fig. 9). In sharp contrast to underdoped materials, the scattering rate shows a temperature dependence over a much broader frequency range, extending at least up to 2000 c m 1. There are indications that the frequency dependence of 1/T(co) becomes superlinear in the strongly overdoped cuprates. The overall magnitude of 1/r(co) is reduced compared to underdoped or even optimally doped materials. The effective

458

D.N. BASOV and T. TIMUSK

"T, E

i Y123, x=6,6 Tc= 0.62 Tma~=58Kc

Tc = 0.73 Tr"ax=67Kc

T=300K

T=300K

I~ 4000

0

I

4000

Bi2212

-

'

I

TY124 c = 82K

I Bi2212 lunderdoped~T=300K __ Tc= 0.69 T;aX=82K T=300K

85K

r

90K

I

'

]

i

E o

,..,2000 150

73K

"

0 4

~E

T=IOK - - k 65K

T=10K

_

3

,~0K

200K 300K

- - ~ * ,

r=10K

_52

-

,~0K _

300K-

2 1 0

1000

1000

1000

0

1000

2000

Wave Number (crn1)

Fig. 7. Optical constants (in-plane) for a series of underdoped high-T~ superconductors.Top row: real part of the optical conductivity.Middle row: frequencydependentscatteringrate. Bottom row: mass renormalization. Experimental data for Y123 and Y124 compounds from Basov et al. (1996); for Bi2212 compounds from Puchkov et al. (1996b). The scatteringrate curves are essentiallytemperature independentabove 1000cm 1but develop a depressionat low temperature and at low frequency.The effectivemass is enhancedat low temperature and at low frequency. ExperimentalresuIts in the normal state at T ~ T~ and in the superconductingstate are remarkably similar.

mass is very close to me. These two results indicate that charge carriers in overdoped cuprates are no longer heavily dressed due to electron-electron and/or electron-boson interactions. Thus the sequence o f figs. 7 - 9 reveals the evolution from a strongly correlated solid with interactions-enhanced effective mass to a weak-interaction material with m* ~ me as exemplified by overdoped T12201. Unfortunately, the experimental results for overdoped crystals are only fragmentary, and we are unaware of any systematic studies of charge dynamics in the overdoped regime besides those depicted in fig. 9. Therefore it remains to be seen if overdoping eventually leads to Fermi-liquid behavior. Based on the existing results for T12201 such a conclusion would be premature. Figure 10 presents a simplified phase diagram based on the experimental results discussed above. The characteristic temperatures T* and Tc are plotted as functions of the carrier density in the CuO2 planes. We have chosen the superfluid density O)ps for the horizontal axis in this particular plot because the magnitude of COps can be determined unambiguously from infrared results (see sect. 5. l). Evaluation o f the free-carrier density

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

"T E

~t TY1'23,x~6.95 ' c= T;a,=93.5K 4000

-

Bi2212 i T=90K C

' t T=3OOK /1 Tl2201 90K _ ~ Tc= 9 0 K

T=9300OK -

T=300K 90K 10K

T=300K90K 10K

459

'

Io 0 4000

T=300K 95K 10K

'~ 2000

0 95K

90K

90K

v

E

0

1000

0

1000

0

1000

Wave Number (cm "1) Fig. 8. Optical constants (in-plane) for a series of optimally doped high-Tc superconductors.Top row: real part of the optical conductivity. Middle row: frequency-dependent scattering rate. Bottom row: mass renormalization. Experimental data for Y123 compounds from Basov et al. (1996); for Bi2212 and T12201 compounds from Puchkov et al. (1996b). Depression of the scattering rate at low frequencies is now observed only in the superconducting state. Also, the spectra of 1/v((~)) acquire some temperature dependence at high frequencies. is difficult since the free-carrier contribution to o(co) is mixed up with the response o f the bound carriers as discussed in sect 4.1. In the underdoped samples there is considerable difference between Tc and T*. The latter can be nearly as high as room temperature. There are indications that in the La214 series, T* can exceed 3 0 0 K (Startseva et al. 1999a). A detailed examination o f this issue will require infrared experiments to be conducted at elevated temperatures. As the doping is increased, T* is suppressed while Tc increases. In the Y 1 2 3 - Y 1 2 4 series, the two boundary lines merge close to optimal doping. But both in Bi2212 and in La214 (Puchkov et al. 1996c, Startseva et al. 1999a), the pseudogap can be seen in overdoped samples and T* > T~ at optimal doping. Combining the phase diagram o f fig. 10 and the experimental data for under- and optimally doped cuprates (figs. 7, 8) it appears that the temperature T* is associated with qualitative changes in the in-plane electrodynamics. indeed, below this cross-over temperature the conductivity spectra show a significant narrowing o f the Drude-like feature and a threshold structure around 500 c m 1. In crystals for which T* >> T~, a further decrease o f temperature below Tc does not lead to significant changes in the conductivity, apart from the formation o f the delta peak at o) = 0 that will be discussed in sect. 5. So far, there is not enough experimental information to extend the

460

D.N. BASOV and T. TIMUSK "T

E

20000

Bi22()1

'T

-

T

c

'

I

I

90K

10000

T=300K 75K 10K

- -

1OK

._>

I

4008

_

t

T=300K 75K 10K

10K

8

T=300K 35K 10K

I

T=300K

90K

v

TI2201 T = 0 . 2 6 T max= 2 3 K C C

Tc= 0.79 T~aX=70K

c

T=300K

o L.)

I

Bi/Pb2212

= 0,89 Tm'X=82K

T=300K 200K 120K 35K 10K

I-

2000

t-,

I-

T=10K 73K

T=IOK

T=10K 35K

- -

300K

300K

3

E

2 i

1 0

[

1000

i

1000

0 Wave

Number

0

1000

2000

(cm -1)

Fig. 9. Optical constants (in-plane) for a series of ouerdoped high-Tc superconductors. Top row: real part of the optical conductivity.Middle row: frequency dependent scattering rate. Bottom row: mass renormalization. Experimental data from Puchkov et al. (1996b). Note that the overall magnitude of 1/r(o)) progressively decreases with doping. phase diagram into the overdoped regime. The charge dynamics in strongly overdoped samples have been studied only for T2201 (Puchkov et al. 1996c), and it remains to be seen whether other systems will behave in a similar fashion in other materials.

4.4. Experimental results." conductivity along the Cu-O chains From the structural point o f view, a distinguishing feature o f the YBa2Cu3Ox and YBa2Cu408 cuprates is the presence of one-dimensional C u - O chains extending along the b-axis of the crystals. The principal role of the Cu O chains is to provide carriers for the CuO2 planes, and in this sense their function is similar to that o f other charge-reservoir layers such as B i - O or T1-O layers in other compounds. Because the chains have a welldefined direction in the crystal it is feasible to study their response through spectroscopic experiments conducted with polarized light. Such experiments were performed as soon as the first untwinned samples became available. The main experimental result is that the reftectivity o f Y123 crystals shows a remarkable anisotropy, the b-axis reflectance being enhanced (Bozovic et al. 1988, Petrov et al. 1988, Koch et al. 1989, Zibold et al. 1990, Schlesinger et al. 1990, Pham et al. 1991, Schfitzmann et al. 1992, S.L. Cooper et al. 1992, Tanner et al. 1992, Basov et al. 1995a). The position o f the plasma minimum in

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

461

300

250

T* ,seudoqa~ s,tate

200

P 150

E

ID

P" 100

50

4x107

doping:

8X107

12;(107

Is~p~,t~.i~ ~,~i~,. ~i~"~c~!'2 i

Fig. 10. Schematic phase diagram of Y123 and Y124 cuprate superconductors. In the underdoped regime, a pseudogap state forms below a temperature T* > T~. The curves for T* and T~ cross at optimal doping where the pseudogap and the superconductivity develop at the same temperature. T* is determined from the c-axis conductivity; tile doping level is determined from the superfluid density ro~ nJm* in the CuO~ planes. From Puchkov et al. (1996a).

reflectance is about 1.2 eV for E 11a and about 2.0 eV for E I I b. The conductivity obtained through KK analysis shows significant enhancemem o f the b-axis spectral weight that can be interpreted in terms o f an additional conductivity channel associated with the chains. The analysis o f the superconducting-state conductivity shows that the superfluid density is also anisotropic (Basov et al. 1995a). In Y124 crystals which contain two C u - O chains in each unit cell, the anisotropy of both the normal-state spectral weight (Bucher et al. 1990, Wachter et al. 1994, Kircher et al. 1993) and of the superfluid density (Basov et al. 1995a) is stronger than in the single-chain Y123. The anisotropy is also observed in the dc conductivity, with Ode along b enhanced compared to the a-axis results (Friedmann et al. 1990, Welp et al. 1990, Rice et al. 1991). While there was considerable scatter in the data for the anisotropy in the samples studied in 1988-1992, more recent results seem to indicate that the ratio ab/cr~ is the same as co2b/cop2~ (Zhang et al. 1994, Basov et al. 1995a). Applying a simple Drude formula to these results, ~rdc = C02/T suggests that the scattering rate is the same along the CuO2 planes and along the chain direction. The chains contribute to dc transport or to the superfluid density only if special care is taken to preserve them from structural disorder in the sample preparation process. Prior to 1991/92, Y123 single crystals were usually grown in Au or A1203 crucibles. Both Au and A1 tend to substitute Cu sites in the C u - O chains. Such contamination does not occur in more recent samples which are usually grown in yttria-stabilized zirconia crucibles. The in-plane response o f the earlier crystals does not differ much from the properties found for the new generation of single crystals. However, the b-axis results are completely different. Rotter et al. (1991) and Schiitzmann et al. (1992) observed a peak at finite frequency in

462

D.N. BASOVand T. TIMUSK

the b-axis conductivity of the Cu-O chains which is characteristic for the response of a 1D metal with localized states. The penetration depth was found to be isotropic in these crystals. No localization peak was observed in more recent samples grown by two different groups (Basov et al. 1995a). The b-axis conductivity of these newer crystals was essentially identical to the a-axis results, except that the spectral weight was enhanced by a factor of 2.3. The same factor appears in the anisotropy of the resistivity and of the penetration depth. 4.5. Experimental results." the interplane conductivity A common feature of all high-Tc superconductors is their layered structure which leads to a strong anisotropy of the electronic properties. While typical values of the in-plane conductivity in most cuprates are of the order of 3000-5000 ~-1 cm-1 at room temperature, the interplane conductivity is much smaller, varying from a maximum of about 400~2 l cm 1 in overdoped Y123 crystals down to 0.01g2 l cm-1 in the most anisotropic Bi2Sr2CaCu2Os compounds. The Y123- and Y124-series compounds are less anisotropic than other materials, which has been attributed to the presence of the Cu-O chains in the crystal structure. However, an estimate of the mean free path along the c-direction, lc, carried out for even the most conductive overdoped samples shows that lc is of the order of the interlayer spacing (Schiitzmann et al. 1994). This is clear evidence that the c-axis transport in cuprates is incoherent. A number of mechanisms have been proposed that would result in an incoherent conductivity spectrum (Kumar et al. 1990, Kumar and Jayannavar 1992, Rojo and Levin 1993, Ioffe et al. 1993, Nyhus et al. 1994, Clarke et al. 1995, Alexandrov et al. 1996). With low electronic background, the c-axis conductivity spectra are dominated by sharp resonances due to infrared active phonons (fig. 4, fig. 11). There are strong indications for the interaction ofphonon modes with the electronic background: (i) the frequency position and the width of phonon peaks is temperature dependent, with anomalies occurring at the onset of superconductivity (Thomsen and Cardona 1989, Litvinchuk et al. 1994); (ii) some of the phonon modes have an asymmetric (Fano) lineshape which is a signature of the coupling between a sharp line and a continuous background (Fano 1961). The line strengths, in contrast, are not anomalous; they are what one expects from latticedynamics models based on neutron scattering (Timusk et al. 1995a). The only exception is the mode at 560 c m 1 involving the apical oxygen atom, which appears to be anomalously strong. It is also the mode with the most asymmetric lineshape. The first step in the analysis of the electronic contribution to the c-axis conductivity is to subtract the phonon part. Usually, the phonons are fitted with Lorentzian oscillators. The high-frequency modes have asymmetric shapes and are best fitted using Fano theory (Burlakov et al. 1989, Homes et al. 1993b). In general, this procedure of separating the conductivity into electronic and phonon channels is quite unambiguous since lattice modes are very narrow and can be easily distinguished from the flat electronic background. The underdoped Y123 and Y124 crystals are exceptions since an anomalously broad phonon-like feature develops at low temperatures (Homes et al. 1993b,

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW ,

800 -

600~. 400 •

/ ~ .~'

,

463

,

YBa~Cu30~, Eli (a)

I

400

0 ~

I

1

300 .

-

1300

1

..... ~m0 K K 1oo K

(b)

-

200

200'

y-6.8

I0 K A

tO0

-"

T -

~ I00

300

(c)

w

y-6.7

|~

~

I

v

>... £--,

/ , , ~--4" ~,-2

gso 0

~

zso

>_

I(d )

l

E-

I

y-6.6

i

'

100-

l/

r,

'(e)

~

'

y-6.5

" ,

200 tOO 0 20O 150 tO0

Z O rD o

0 40O

5O 0 200 i50

ii

100

YB.,c~,o..

50 0

0

o

2oo

4oo

6oo

Ioo

0

200

FREQUENCY

400

600

( c m -~)

Fig. i 1. Optical conductivity for YB½Cu3Or+y for radiation polarized along the c-axis as a function of oxygen doping. Left panel: results by Tajima et al. (1998); right panel: results by Homes et al. (1995a). In the highly doped materials, a gap-like depression at co < 500cm -I opens up below T~, but at 10K, the conductivity remains finite down to the lowest accessible frequencies. In the x = 0.85 material, the conductivity is clearly non-metallic, a trend which resolves itself into a pseudogap for x < 0.70.

Basov et al. 1994a, H a u f f et al. 1996). S u m - r u l e arguments suggest that this m o d e is phonon-related: the spectral weight o f this m o d e grows at low temperature at the expense o f other p h o n o n resonances. However, the halfwidth o f this feature exceeds 50 c m 1, which is atypical for p h o n o n s in crystalline solids. A n unexpected result is that this m o d e disappears w h e n impurities are added in Y123 materials (Hauff et al, 1996). The frequency o f this m o d e is proportional to To (Homes et al. 1993b) similar to the m a g n e t i c p e a k seen by n e u t r o n scattering ( R o s s a t - M i g n o d et al. 1991, M o o k et al. 1993, Fong et al. 1995), b u t the infrared line is at somewhat higher frequency. Possible caveats o f coupling

464

D.N. BASOVand T. TIMUSK

between lattice modes and electronic conductivity were discussed by Grtininger et al. (2000), Munzar et al. (1999) and Bernhard et al. (2000). The electronic background in the interplane conductivity shows a rich variety of behaviors as the system is doped with charge carriers. For a given series, higher conductivity correlates with higher doping levels. S i . Cooper et al. (1993b) have suggested the following form for the c-axis response of Y123:

a(~(~o) o( tc2o1 + co2T~'

(13)

where 1/re is the c-axis scattering rate and to0 is the interbilayer hopping rate. For materials with an open Fermi surface in the c-direction, the interbilayer hopping is expected to scale linearly with the c-axis plasma frequency (Kwak 1982). Within this phenomenological model, the interbilayer hopping rate changes from 40-50 meV in the optimally doped Y123 samples down to 5 meV in the underdoped 60 K phase. However, it should be pointed out that this model assumes coherent interlayer transport, and since the Drude width (160 meV, Schfitzmann et al. 1994) exceeds the interlayer hopping there cannot be any coherent bands in the c-direction, and there is a complete breakdown of the Fermiliquid picture these arguments rest on. The slope of the dc resistivity P,tc = 1/~rdo changes from metallic (dp/dT > 0) in the optimally doped and overdoped samples to "semiconducting" (dp/dT < 0) in underdoped compounds (Ito et al. 1993, Nakamura and Uchida 1993). The cause of this peculiar behavior was unraveled by Homes et al. (1993a) who analyzed the conductivity spectra for YBa2Cu3Ox crystals with x = 6.95 and x = 6.7. The room-temperature spectra measured for the optimally doped crystal showed a continuous background which was nearly independent of frequency in both compounds, in agreement with the earlier data (Koch et al. 1989, S.L. Cooper et al. 1993b). At lower temperature, the x = 6.95 crystal reveals weak metallic character: the low-frequency part of the conductivity shows weak Drude-like behavior and (~1(~o ~ 0) increases. This is in an agreement with the positive dp/dT found in the resistivity measurements. But it should be noted that there is a very large residual resistance giving a mean free path for impurity scattering of the order of a lattice spacing. A completely different temperature dependence was observed in the spectra of the underdoped x - 6.7 sample. In this crystal the spectra of ol(~o) at T < 140-160K reveal a low-frequency pseudogap: the region in the frequency dependence located at co < 200-300cm I where Crl(CO) is suppressed but remains finite. Judging from the ~) position of the step-like structure in the spectra of Crl(O)), the magnitude of the pseudogap remains unchanged with temperature (~250cma). The opening of the pseudogap is associated with the transfer of the spectral weight from the low-energy part of the spectra to higher energies. This accounts for the increase of the dc resistivity, and in fact there is good agreement between the T-dependence of resistivity measured directly and that estimated estimated from o1(co -+ 0) (Homes et al. 1995a). An interesting result is that no qualitative changes are found in the conductivity spectra as underdoped

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

465

samples undergo the superconducting transition. That is in contrast with the behavior of conventional superconductors where the ac conductivity is zero below the gap frequency. A pseudogap similar to that in fig. 11 was observed in stoichiometric YBa2Cu408 crystals which are intrinsically underdoped (Basov et al. 1994a). Experiments by Homes et al. (1995a) and by Tajima et al. (1998) explored the evolution of a pseudogap with doping in the YBa2Cu3Ox series. Their major finding is that the magnitude of the pseudogap is independent of doping but the temperature T* at which the pseudogap structure appears in the spectra seems to increase as x is reduced. The dependence of T* on carrier density in the CuO2 planes is shown in fig. 10. In the samples with x = 6.5-6.6, the pseudogap is found already at 300 K. It has been noted that the c-axis pseudogap is related to other anomalies observed in underdoped samples in the normal state. Homes et al. (1993a)have shown that the T-dependence of the conductivity in the pseudogap region agrees with the behavior of the Knight shift in YBa2Cu3Ox. A similar conclusion was drawn from the data for Y124 compounds (Basov et al. 1994a). Basov et al. (1996) have shown that the pseudogap in the interplane conductivity develops in the same temperature range and in the same doping regime as the suppression of the in-plane scattering rate. Previously, resistivity measurements conducted for currents along the ab-plane and along the c-axis demonstrated that suppression of the in-plane resistivity both in Y123 and in La214 crystals coincides with the cross-over to "semiconducting" behavior observed for the interplane currents (Nakamura and Uchida 1993, Takenaka and Uchida 1994). These results establish a link between the anomalies in the charge dynamics observed in the c-axis and in the ab-plane properties. A different behavior was observed by Schtitzmann et al. (1994) in the c-axis response of lightly overdoped Y123 sample. The absolute value of the conductivity at co ---+ 0 increased, and the spectra clearly show a Drude-like peak at co = 0 which is characteristic for a metallic system. The authors analyzed their data within the extended Drude model (fig. 12). The results of this analysis reveal certain similarities with the in-plane properties of the Y123 series. Both the scattering rate 1/•(co) and the effective mass m*(e)) show a pronounced frequency dependence. Comparing interplane spectra with the data for the in-plane response (fig. 8), one finds that the effective masses probed along the CuO2 planes and along the c-direction do not differ by more than a factor of 2. This contradicts the predictions of the band-structure calculations where the anisotropy is found to be as high as 10 in YBa2Cu307 (Pickett et al. 1990). Thus one is forced to conclude that the anisotropy of the dc resistivity, which is about 10 in YBa2Cu307, cannot be completely accounted for by band-structure effects but can be connected with the peculiarities of the interplane tunneling or hopping. A naive interpretation of the interplane conductivity in terms of the Drude model suggests a strong enhancement of the scattering rate in the c-direction. Anisotropic damping is an interesting phenomenon that has been predicted as the result of spin-charge separation (Clarke et al. 1995, Chakravarty and Anderson 1994). However, one should strike a note of caution here about the applicability of the frequency-dependent scattering-rate formalism in a situation where the mean free path is clearly smaller than the lattice spacing and the conduction is incoherent.

466

D.N. BASOV and T. TIMUSK

300Ci \~..

300~,,

'

'

T

T

F

=._20O0 :>-, 1000

3h~

/lOOK

,

o

o

lOOO Frequency (crn"I)

2000

Fig. 12. The results of the extended-Drude analysis of the optical conductivity of a "fully oxygenated" Yl23 single crystal measured with the polarization along the c-axis (Tajima et al. 1998). The crystal is likely to be overdoped as evidenced by the reduced T~= 89 K and enhanced conductivity in the limit of co ~ 0. Note that the enhancement of the effective mass m*/m e does not exceed -4 which is close to the in-plane values (figs. 8, 9).

The c-axis response o f La214 crystals in a broad range o f doping regimes has been investigated by Uchida et al. (1996). In the far infrared range the electronic background in the conductivity is nearly frequency independent. The main difference from the Y123 or Y124 series is that the conductivity background is much weaker, and even in overdoped samples oc(co --~ 0) does not exceed 40 ~2 1 cm 1. Underdoped samples show a suppression o f the conductivity in the low-frequency part. This is concomitant to the "semiconducting" behavior of the resistivity found in underdoped La214 (Nakamura and Uchida 1993). However, the energy scale that can be associated with the temperature dependence of the resistivity and especially the frequency dependence o f crj(to) is very different in the Y123 and La214 families. Basov et al. (1995b) have shown that the spectral weight removed from the low-frequency part is recovered at co exceeding 0.4 eV This suggests that a much broader energy scale is associated with the pseudogap in La214 materials. Batlogg et al. (1994) and Batlogg and Emery (1996) have presented arguments in favor of the existence o f a temperature scale of the order of 1000 K which is relevant to the transport properties o f high-Tc materials. From an analysis o f the anisotropy in the conductivity spectra o f Lal.ssSr0.lsCuO4, van der Marel and Kim (1995) concluded that mc/mab does not exceed 5 in this compound. A larger anisotropy was inferred by Henn et al. (1996) from an analysis o f ellipsometric data for Lal.87Sr0.13CuO4. This latter analysis is based on the extended Drude model. Again one must question the applicability o f this model to the c-axis response o f this compound since al ((o) at (~ -+ 0 is smaller than the minimum metallic conductivity.

1NFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

467

4.6. Comparison of the optical data with transport measurements and other

spectroscopies In an attempt to understand the fundamental physics behind the electromagnetic response of cuprates it will be instructive to discuss connections to the results of other spectroscopic studies and, first of all, to transport measurements. It was realized early on that the linear Ji'equency dependence of the scattering rate found in the response of the CuO2 planes may be related to the linear temperature dependence of the ab-plane dc resistivity Pdc = m/ne 2"c. These two effects occur when quasiparticles scatter from a hypothetical bosonic spectrum which is flat over a frequency scale T < co < coo (Littlewood and Varma 1992). Using this model, known as the Marginal Fermi Liquid (MFL) model, it was possible to rationalize the results obtained with photoemission, Raman or tunneling spectroscopies. Ruvalds (1996) discussed this behaviour as a consequence of Fermi surface nesting. However, the linear laws in 1/r(co) and in dp/dT are not universal, and a more complicated response is observed in underdoped samples. Specifically, a broad range of physical probes show that below a characteristic temperature T*, which can significantly exceed Tc, the response of the underdoped cuprates can be understood if one assumes there is a partial gap or a "pseudogap" in the spectrum of low-energy excitations. Originally, the gap-like behavior in the normal-state properties was observed in the uniform spin susceptibility probed with nuclear magnetic resonance (NMR) measurements of the Knight shift (Warren et al. 1989, Yoshinari et al. 1990, Alloul et al. 1991). In ordinary metals and in the optimally doped cuprates, the Knight shift is temperature independent at T > T~ but drops rapidly below the superconducting transition due to the formation of (singlet) Cooper pairs. However, in underdoped cuprates the Knight shift is suppressed already at temperatures well above To. This led Warren et al. to the conclusion that in underdoped samples spin pairing may take place well above the bulk superconducting transition at To, thus producing a normal-state energy gap, referred to as a "spin gap". In the temperature range in the vicinity of T*, the ab-plane resistivity in underdoped cuprates shows a deviation from the linear behavior (Bucher et al. 1993, Batlogg et al. 1994, Ito et al. 1993). Both in Y123 and in La214 materials the resistivity reveals a steeper than linear slope at T < T*, with a cross-over to linear behavior at T > T*. Bucher et al. (1993) and Batlogg et al. (1994) have shown that anomalies in resistivity and in NMR data occur in the same temperature range, and suggested that the two effects are interrelated. I f the scattering mechanism responsible for the linear temperature dependence of p,h(T) involves spin fluctuations, then the spin gap seen in NMR below T* would naturally account for the drop of pal,(T) below T* as well. Suppression of the spin susceptibility is also seen in neutron scattering experiments (Rossat-Mignod et al. 1991, Tranquada et al. 1992) as well as by specific-heat measurements on underdoped Y123 (Loram et al. 1994). However, in the latter work it was shown that there is a large decrease in entropy below a temperature closely related to T*, which cannot be accounted for by assuming a gap only in the spin degrees of freedom. Recently, angle-resolved photoemission spectroscopy (ARPES) for Bi2212 showed that the electronic density of states of underdoped cuprates reveals a normal-state gap-

468

D.N. BASOV and T. TIMUSK

like depression (Marshall et al. 1996, Loeser et al. 1996, Ding et al. 1996, 1997). The momentum dependence of this normal state gap resembles that of the dx2_/gap observed for T < Tc (Loeser et al. 1996, Ding et al. 1997). This, and the fact that no significant changes are observed upon crossing into the superconducting regime, have led to the suggestion that the normal-state gap may be a precursor of the superconducting gap. As the doping level is increased to near- and above optimal, the normal-state gap-like feature seen in ARPES spectra disappears. In the optical conductivity, the pseudogap state reveals itself in two distinct ways: (i) the in-plane scattering rate at co < 700cm 1 is reduced which inevitably leads to reduction of the in-plane resistivity (see discussion in sect. 4.3); (ii) the interplane conductivity is reduced at co < 250 c m I which accounts for semiconducting behavior of the c-axis resistivity (see discussion in sect 4.5). Despite the difference in the energy scales seen in the in-plane and interplane transport, these two effect are clearly connected (Takenaka and Uchida 1994, Basov et al. 1996). Recently, a gap in the normal state has also been seen by vacuum tunneling spectroscopy in underdoped cuprates (Tao et al. 1997, Renner et al. 1998, Krasnov et al. 2000). Renner et al. (1998) showed that in Bi2212, the pseudogap above T~ scales with the superconducting gap, and its width is temperature independent; i.e., the gap does not close as the temperature is raised. The maximum gap, assuming a d-wave gap, at optimal doping in Bi2212, A0 = 41.5 meV, is consistent with other spectroscopies. The gap starts with z~0 = 44 meV for the underdoped sample with T~ = 83 K and decreases with doping to A0 = 21 meV in an overdoped sample with Tc = 56 K. Interestingly, Renner et al. (1998) find that even in an overdoped sample with Tc = 74.3 K there is a pseudogap in the normal state although it is very weak already at 89 K. These observations are consistent with the phase diagram shown in fig. 10 where the temperature T* drops rapidly with doping, but in the case of Bi2212, at least, the temperature scales T* and Tc do not cross at optimal doping but at a doping level that is substantially higher. This is also seen in the ab-plane scattering rate for Bi2212 (Puchkov et al. 1996a). Raman spectroscopy is also capable of probing the pseudogap. In conventional superconductors, the Raman spectrum only shows a peak at the pair-breaking frequency 2A (Abrikosov and Fal'kovskii 1961). In the clean limit no other absorption is expected apart from a very low frequency band from free electrons, but in the cuprates, because of the strong incoherent scattering (Shastry and Shraiman 1990), a broad continuum appears that gives rise to a constant Raman background up to 2 eV (Varma et al. 1989). A depression in this continuum at low frequency has widely been interpreted as evidence of the superconducting gap, but its presence in the normal state in underdoped samples originally cast doubt on this interpretation (Slakey et al. 1990). It is now becoming clear that this depression may well be associated with the pseudogap and that the Raman spectrum has parallels with the ab-plane scattering-rate spectrum (Blumberg et al. 1997). At optimal doping, Raman spectra of Big and B2g symmetry are quite different, with Blg spectra emphasizing processes involving states in the (Jr, 0) direction of the maximum gap, whereas the B2g spectra involve states near (Ic, Jr) in the region of the nodes of

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

469

the dx2 y2 function. Fits of a simple theory to the data for optimally doped Bi2212 in the superconducting state give a maximum gap value of A0 = 280 cm -~ (34.7 meV) in reasonable agreement with the gap value estimated from other probes (Hackl et al. 1996). Recent Raman studies of underdoped cuprates are being interpreted in terms of a pseudogap (Blumberg et al. 1997, Nemetschek et al. 1997). The opening of the pseudogap leads to a suppression of the electronic scattering at co < 500 c m I with the simultaneous formation of a bound state at co ~ 600 c m 1. These two effects are reminiscent of what is observed in Raman scattering in optimally doped cuprates as they undergo the superconducting transition. Blumberg et al. interpret the sharp bound state as being due to pair formation that takes place in underdoped crystals at a temperature significantly exceeding T0. There have been attempts to analyse the low-frequency Raman scattering in terms of the carrier lifetime F(£2). In the simple Drude picture of the Raman continuum scattering in the normal state, the Raman cross-section is proportional to co~F, where F is the scattering rate. In this model a depressed scattering rate in the pseudogap state would lead to an increased slope for the low-frequency scattering. Naeini et al. (1998) investigated overdoped La2_xSrxCuO4 using this analysis, finding no pseudogap for a sample with x = 0.22, in contrast to Startseva et al. (1999b) who found a pseudogap in the ab-plane conductivity scattering rate in the same material. Nemetschek et al. (1997) do find a depression in the Raman scattering of both Bi2212 and Y123 in the normal state below 700 cm -~ and 200 K in agreement with the ab-plane optical scattering rate data. The anomalies observed in tmderdoped samples in the dc resistivity, the optical conductivity, neutron scattering, NMR and ARPES, Raman and tunneling spectroscopies are summarized in table 1. The similarities in the evolution of these anomalies with temperature and doping suggests they have a common physical origin. However, a closer look at experiments reveals several potential discrepancies. For example, the energy scale of the pseudogap varies substantially from one experimental probe to another. Also, there does not seem to be any simple relationship between the energy scale set by the magnitude of T* and the frequency position of features seen in a variety of spectroscopies. The challenge for any theory is to give an account for all these distinctions and to reconcile (seemingly) conflicting results. From table 1 it follows that the opening of the pseudogap in the electronic DOS as seen in ARPES and tunneling experiments is accompanied by an enhancement of the dc conductivity. It may seem that the higher dc conductivity contradicts the observation of a gap. Indeed, the opening of a gap at EF reduces the spectral weight available for conduction. However, in a system where the scattering excitations themselves are gapped, 1/r collapses and this improves the dc conductivity. Chromium is a canonical example of a remarkably good metal where a gap corrupts most of the electronic states at EF (for a comprehensive review, see Fawcett 1988). This partial gap is due to the spindensity-wave transition at TsDw = 312 K. However, some fragments of the Fermi surface remain unaffected by the transition. The electrical resistivity shows a weak increase at T ~ Ts~w which is followed by a fast drop o f p ( T ) with decreasing temperature. A similar mechanism of pseudogap formation in cuprates was discussed by Kampf and Scbrieffer

D.N. BASOV and T. TIMUSK

470

Table 1 Transport and spectroscopic probes of the pseudogap state in cuprates Experiment

Pseudogap at T < T*

System

Characteristic Torso

Reference

a-axis resistivity

faster than linear decrease ofp(T)

o'c,(~)

suppression of the scattering rate

c-axis resistivity

"semiconducting" up-turn

ac(w)

suppressed in FIR

ARPES

incomplete gap at E~, d-wave symmetry

Raman

suppressed electronic background gap with states at E F

La214 Y123 Y124 Bi2212 La214 Y123 Y124 Bi2212 Bi2201 Iig1223 La214 Y123 Bi2212 La214 Y123 Y124 Bi2212 Bi2212 Bi2201 Bi2212 Y123 Bi2212

300 800K 120-300K 140 180K 120-220K 700cm i 600cm -1 600cm ~ 600cm 1 600cm ~ 1000c m 1 500-800K 120-300K 120-300K 2000cm 1 <250cm 1 <250cm ~ <200cm I <160 cm-I <120cm ~ <600 cm ~ <700 cm z <700cm i <560 c m j

Baflogg et al. (1994) Ito et al. (1993) Bucher et al. (1993) Watauabe et al. (1997) Startseva et al. (1999a,b) Basov et al. (1996) Basov et al. (1996) Puchkov et al. (1996b) Weber et al. (1997) McGuire et al. (2000) Nakamura and Uchida (1993) Takenaka and Uchida (1994) Watanabe et al. (1997) Basov et al. (1995b) Homes et al. (1993b) Basov et al. (1994a) Loeser et al. (1996) Ding et al. (1997) Harris et al. (1997) Blumberg et al. (1997) Nemetschek et al. (1997) Renner et al. (1998) Ka'asnovet al. (2000)

Tunneling

(1990). A c c o r d i n g to the A R P E S experiments, the p s e u d o g a p in cuprates is likely to have d-wave symmetry. Thus, the regions close to the nodes are capable o f maintaining substantial conductivity. T h e dc c o n d u c t i v i t y will be d e t e r m i n e d by a c o m p e t i t i o n b e t w e e n the reduction o f the D O S and the collapse o f the scattering rate. E x p e r i m e n t a l e v i d e n c e suggests (figs. 7, 8) that the decline o f 1/T prevails, thus leading to an increase o f the dc conductivity in the p s e u d o g a p state. A striking e x a m p l e o f this effect is the conductivity p e a k o b s e r v e d in m i c r o w a v e spectroscopy in the superconducting state o f optimally d o p e d cuprates ( B o n n et al. 1992). 4.7. Theoretical context

A resonance in the scattering spectrum. In searching for possible connections b e t w e e n the anomalies o b s e r v e d in the p s e u d o g a p state (table 1), it m a y be instructive to p e r f o r m one m o r e step in the s i n g l e - c o m p o n e n t analysis o f the optical conductivity, that is, to extract the f r e q u e n c y d e p e n d e n c e o f A(g2) f r o m the experimental data. W h i l e it is difficult to

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

471

invert eq. (12), the gross features of the A(Q) spectrum can be inferred from comparison of the data shown in figs. 7-9 with the model calculations described in sect. 4.2. The principal result of such a comparison is that eq. (12) can account for the experimental data only if the scattering spectrum A(Q) is strongly dependent on temperature and doping. Indeed, the spectral weight in A(Q) has to be suppressed at low frequencies in order to account for the development of a threshold feature in the 1/T(o)) spectra for underdoped materials. Moreover, in optimally doped samples, and in particular in YBa2Cu306.95, the threshold develops instantly at T < Tc, consistent with the observation of an abrupt drop of the scattering rate in the superconducting state by means of microwave measurements (Bonn et al. 1992). Finally, in overdoped materials, the absolute value of the scattering rate through the entire energy range is strongly diminished compared to compounds with lower doping. Within the electron-boson formalism this implies that the total spectral weight of A(Q) is strongly reduced since 1/r o( J0°~ dQA(f2). A simple suppression of A(£2) at low frequencies is not sufficient to model the weak T-dependence at e) > 700 cm t found in underdoped and optimally doped samples. To account for this effect the spectral weight of the scattering spectrum has to be recovered right above the threshold structure in 1/r(c0). This was shown by Puchkov et al. (1996a) using the inversion formula proposed by Marsiglio and Carbotte (1997). The main difficulty with the inversion procedure is that differentiation greatly amplifies any noise in the experimental data. While it is not possible to find fine structure in A(g2) with this analysis, it suggests that a resonance at ~o ~ 500 cm -1 is formed in the pseudogap state. In the optimally doped samples, 1/T(~o) becomes "gapped" only at T < T~ and, therefore, the resonance is expected to occur only in the superconducting state. Thus, in principle, there exists an alternative explanation of many of the experiments listed in table 1 in terms of a gap in the scattering spectrum. This scenario does not necessarily have to be contrasted with the one in which the pseudogap is viewed as a suppression of the DOS. Indeed, if the scattering mechanism is of electronic nature, a distinction between the former and the latter approaches is difficult to make. A resonance in the scattering spectrum defines the electronic self-energy 2J(co), and thus would determine the behavior of the spectral function A(k, e)) according to eq. (10). It has been suggested that photoemission experiments can be interpreted in terms of the one-electron spectral function and, therefore, the photoemission lineshape can be calculated based on the knowledge of the self-energy (Mahan 1980, Kampf and Schrieffer 1990). This, in principle, allows one to check the consistency between infrared and photoemission results on a semi-quantitative level. Recently, Shen and Schrieffer (1997) and Norman et al. (1997) suggested that the photoemission spectra are consistent with quasiparticle coupling to a collective excitation centered near Q = (~c, :v). Norman et al. (1997) have argued that the "peak-dip" structure of the photoemission spectra seen in the superconducting state of the optimally doped samples can be accounted for if one assumes that this is a collective excitation that develops right below Tc. Extending their arguments to tmderdoped materials, one can suggest that the development of this collective mode in the underdoped regime is expected already in the normal state at T < T*. Thus one can conclude that both the photoemission lineshape and the ab-plane optical conductivity

472

D.N. BASOVand T. TIMUSK

of cuprates can be described as the result of the interaction of quasiparticles with the collective mode which shows a remarkable temperature dependence. If this collective mode is related to the spin degree of freedom, this would reconcile numerous correlations between spin and charge responses observed in the cuprates. One of the likely candidates for such a collective mode is an anomalous magnetic peak at 41 meV that has been studied intensively over the last several years with neutron scattering (Rossat-Mignod et al. 1991, Mook et al. 1993, Fong et al. 1995, 1997). Evidence for coupling of charge carriers to a neutron mode was reported by Carbotte et al. (1999). Since many of the pseudogaprelated features are sensitive to the presence of impurities (in particular, Zn) it will be interesting to explore if the 41 meV resonance is found in Zn-doped Y123 crystals. Spin-charge separation. As discussed above, the suppression of the scattering rate at T < T* means that a gap or a pseudogap may open in the spectrum of excitations that scatter the conducting carriers. A close correspondence between charge properties probed with resistivity or in-plane optical conductivity and spin properties probed with NMR or with neutrons suggests that the primary source of scattering is related to the spin degree of freedom. This is what one expects within the model of spin-charge separation (Anderson 1987, Lee and Nagaosa 1992, Ioffe et al. 1994, Altshuller et al. 1996, Laughlin 1996, Millis and Monien 1996). The principal argument behind these ideas comes from NMR evidence for pairing of neutral fermions into singlets. This is to be distinguished from superconducting pairing since no precursors to superconductivity are observed in the pseudogap state: a Meissner effect is not observed in the pseudogap state and there is no paraconductivity due to superconducting fluctuations. This implies spin-charge separation where the in-plane carrier conductivity is due to holons but scattering is due to exchange of spinons. The phase diagram obtained by Lee and Nagaosa (1992) is very similar to that depicted in fig. 10. Thus, the spin-charge separation picture is consistent with the experimental situation on the pseudogap state, at least in materials where the maximum Tc coincides with the disappearance of the pseudogap. Striped phases. An alternative suggestion is that cuprates are intrinsically inhomogeneous and form "charged stripes" (Emery et al. 1997 and references therein). An interaction between hole-rich stripes and hole-free antiferromagnetic regions is a form of magnetic proximity effect which is considered as a likely cause of high-Tc superconductivity. Within this picture, the mobile holes on an individual stripe acquire a spin gap by hopping between the stripe and its antiferromagnetic surrounding. This is a magnetic analog of the superconducting proximity effect. Recently Salkola et al. (1996) have been able to explain the lineshape observed in photoemission experiments using this model. The theory predicts two cross-over temperatures: the first is related to the formation of stripes, and the second is where the pairing (spin-gap) behavior emerges within a stripe. Then the pseudogap can be viewed as the result of superconducting pairs which are preformed already in the normal state (at T < T*) (Uemura 1996). However, no bulk superconductivity is observed until these pairs gain global coherence, which occurs only below Tc. Emery and Kivelson (1995b) emphasized the role of phase fluctuations in the pseudogap state. A phase diagram with two cross-over temperatures has also been obtained recently by Zhang (1997), who suggests a unified approach towards

INFRARED PROPERTIES OF HTSC: EXPERIMENTALOVERVIEW

473

superconductivity and magnetism based on a 5-dimensional order parameter which contains both antiferromagnetic and superconducting components. A different view of the experimental situation is formulated within the so-called nearly antiferromagnetic Fermi-liquid model (NAFLM) (Pines 1995, Chubukov et al. 1996). Within this approach, there is no spin-charge separation. One assumes Fermi-liquid behavior at all doping levels. The anomalous spin and charge behavior seen in the normal-state properties is believed to arise from a magnetic interaction between planar quasiparticles, which reflects the closeness to antiferromagnetism in the phase diagram. The evolution of the Fermi surface with doping is accompanied by the development of a precursor to a spin-density-wave state which gives rise to a pseudogap at low temperatures (Chubukov et al. 1996). Another related model is the spin-bag mechanism of K a m p f and Schrieffer (1990) discussed above. Finally, it should be pointed out that some of the intriguing features of the normal-state behavior of cuprates can be explained starting from the premise that the ground state is metallic but the layered structure is responsible for the complicated tunneling effects between the CuO2 planes (Rojo and Levin 1993, Abrikosov 1995, 1996, Atkinson and Carbotte 1997, Zha et al. 1996). These models are very successful in explaining the behavior of the Y123 materials where many of the effects observed in the optical conductivity and in the penetration depth can be ascribed to the interband transitions involving CuO2 planes and one-dimensional C u - O chains (Atkinson and Carbotte 1995).

5. Electrodynamics in the superconducting state

5.1. Searching Jbr the superconducting energy gap Within the BCS model, the ground state of a superconductor is characterized by an energy gap 2A (Bardeen, Cooper and Schrieffer 1957). The first spectroscopic evidence for the existence of the energy gap in conventional superconductors as well as its correct magnitude was provided by Glover and Tinkham (1956) using far-infrared spectroscopy. These pioneering experiments preceded the appearance of the BCS theory but were soon found to be in excellent agreement with the BCS electrodynamics (Mattis and Bardeen 1958). Spectroscopic measurements of the energy gap are based on the fundamental fact that absorption of radiation with frequency less than the magnitude of the gap is prohibited in a superconductor, and therefore the dissipative (real) part of the conductivity vanishes for co < 2A. A clear threshold in the conductivity spectra at co = 2A is found only in dirty superconductors where 1/T > 2A. In clean materials where 1/T << 2A the conductivity in the gap region is small even in the normal state, and unlike other spectroscopies where a sharp feature can be seen at the gap energy, there is little change in the overall optical properties at the gap frequency in a clean superconductor. A detailed discussion of this issue can be found in a review article by Timusk and Tanner (1989).

474

D.N. BASOV and T. T I M U S K 12000

[

I

YBa2Cu306

I

1

I

I

95 ,30'

*

* * °

; T C::

"7

8000

-

i

T = 95 K

%

10 K

~

,3.8 G H z

== 20

* " 0

•

%"

...".-"

lo ':

34.8 GHz '

•

20

*

"'"-. -.-. "°",,

'

1

40

60

,

: %=° I 80

,=~ 100

4000

0 - 0

I

I 500

t

I

I

I

1000

Wavenumber,

1500

cm

I 2000

-1

Fig. 13. Real part o f the in-plane optical conductivity of a twinned single crystal of YBazCu30695 at T ~ T~ and at T = 10 K (fi'om Basov et al. 1994a). The inset shows the temperature dependence of the real part of the conductivity at co = 34.8 GHz = 0 . 9 9 c m 1 and at co = 3.8 GHz = 0.11 cm ~, obtained for the crystal from the same source with microwave methods (from Bonn et al. 1993).

Infrared spectroscopy is a powerful technique for the study of low-lying states in superconductors, and the number of papers devoted to this subject in the high-To materials is large. In order to isolate the essential features of the superconducting-state response it might be instructive to begin with the ab-plane conductivity of a twinned YBa2Cu306.95 crystal, which is to date the best studied material. In the superconducting state, the conductivity is suppressed compared to the spectrum measured at T ~ T~ over a broad energy range extending at least up to 1000 cm -1 (fig. 13). The conductivity shows a broad minimum at co ~ 500 cm 1 which was often confused with the energy gap. However, there is no evidence whatsoever to assign this feature to the superconducting gap. This is because ol (co) remains finite down to the lowest frequencies in the infrared and microwave frequency regions (Pham et al. 1990, 1991, Renk et al. 1991, Miller et al. 1993, 1995, Basov et al. 1994b). Also, as first pointed out by Reedyk et al. (1988), and subsequently by a number of authors (Tanner and Timusk 1992), in the underdoped samples the minimum of the conductivity is observed already in the normal state, and its frequency position does not scale with Tc (Orenstein et al. 1990). Finally, in the clean limit that applies to high-temperature superconductors (Bonn et al. 1992), no feature is expected in the conductivity at the gap frequency. Basov et al. (1996) have shown that the threshold of

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

475

absorption and the minimum in the conductivity originate from the suppression of the scattering rate at T < T*. Its relation to the superconducting gap is still a matter of theoretical interpretation. While it is difficult to extend infrared measurements below 30-40 cm l, microwave experiments (Bonn et al. 1993) performed on crystals from the same source clearly show the existence of a narrow quasiparticle mode with width 1/~ ~ 1-3 cm -1 at low temperatures. This is consistent with the vanishing of 1/r(~o --~ 0) obtained from the one-component analysis of the infrared conductivity (fig. 8). The width and the spectral weight of the narrow mode in ol (~o) are strongly temperature dependent. The competition between the suppression of the scattering rate in the superconducting state and the reduction of the spectral weight of the normal fluid (due to pair formation) produces a peak in the T-dependence of al(o)) at T ~ 3 0 - 4 0 K (see inset of fig. 13). The residual absorption in the superconducting state at T << T~ is a well-established result. This absorption is found in all cuprates for which accurate data exist. The finite dissipation in the FIR region is remarkably well documented through direct absorption measurements using bolometric techniques (Pham et al. 1991, Miller et al. 1993). In particular, residual absorption was found in the measurements performed for untwinned Y123 and Y124 crystals for the polarization of the E vector along the a-axis (Pham et al. 1991, Schfitzmann et al. 1992, Basov et al. 1995a, Kircher et al. 1994, Holmes and Richards 1997). These latter experiments ruled out earlier suggestions that the residual losses are solely connected with the C u - O chains. Indeed, in a-axis measurements, where the chains are normal to the polarization of the electromagnetic radiation, the chains do not contribute to the optical conductivity but al(CO) was still found to be finite down to the lowest accessible frequencies. Moreover, Timusk et al. (1995b) have shown that the results obtained in three different groups for the a-axis conductivity are in good quantitative agreement with one another. Since there is almost no sample-to-sample variation in the a-axis response of Y123 it is difficult to assign residual absorption to sample imperfections. Chain-free Pb2 Sr2ReCu3 08, Bi2212 and T12201 crystals also show finite FIR conductivity in the superconducting state (Reedyk 1992, Puchkov et al. 1995, 1996c, Barowski et al. 1996). Somal et al. (1996) used a technique of reflectance spectroscopy at grazing angle of incidence to examine the FIR conductivity of Lalss Sr0.15CuO4, The advantage of grazingincidence measurements is that they significantly improve the experimental accuracy in determination of the optical constants for highly reflective substances. The results of these measurements, presented in fig. 14, convincingly show a residual conductivity in this material. At the time of writing it still remains unclear whether or not the ab-plane response of the electron-doped superconductor Nd2_yCeyCuO4 is also characterized by residual losses. Microwave measurements seem to indicate conventional electrodynamic response consistent with an s-wave gap (Wu et al. 1993). But this assertion was not supported by a FIR transmission experiment which did not show any of the expected features of the s-wave BCS superconductor (Choi et al. 1996). The conclusion of an isotropic BCS gap with a magnitude between 4.1 and 4.9kTc (Stadlober et al. 1995) was inferred from Raman

476

D.N. BASOV and T. TIMUSK

1.0 •

.

1.0

L ~

,

,'~ 0.5 i- ~ - " ~ " ~

m

o.9 ~0

rr

o 101 10 2 10 3

•N - 0.8 C9

0.7 5000

E .po o9

~..

.<.%4ooo

6

3000

1-~_~.

4

0

20

40

"O =

O

O

2

F 0

10 20 30 Photon Energy (meV)

40

Fig. 14• Upper panel: ab-plane reflectivity of Lal.85Sr0.15CuO4 at 80 ° angle of incidence at 4 K (solid curve), 20K (dashed), 27K (dotted), 40K (dash-double-dotted) and 50K (chained). Inset: reflectivity at normal angle of incidence for 40K (solid) and 300K (dashed). Lower panel: optical conductivity at 50K (inverted solid triangles), 40 K (circles), 27 K (lozenges), 20 K (solid triangles) and 4 K (squares) obtained from the reflectivities using Kramers-Kronig relations. Inset: skin depth for 4 K (solid) and 20 K (dashed). The solid dots are the penetration depth obtained from the conductivity sum rule.

measurements. In contrast, Choi et al. (1996) concluded that the electromagnetic response reveals certain similarities with the Y123 series which by no means can be classified as an s-wave system. J.R. Cooper (1996) suggested that Nd2._yCeyCuO4 can mimic the behavior of conventional superconductors because Nd acts as a magnetic impurity in this material• The controversy regarding an s-wave gap in electron-doped superconductors can be resolved by measurements of the optical constants in the superconducting state in the energy gap region (2ABcs ~ 50cm-~). This is a demanding experiment, and the only available set of conductivity data does not extend below 50 cm -~ (Homes et al. 1997)• The residual absorption found in the vast majority of cuprates is not consistent with s-wave symmetry of the superconducting order parameter• Since the DOS inside the gap region of an s-wave superconductor is exactly zero at T = 0, the dissipative part of the conductivity must vanish for co < 2A, in clear contrast with the experimental data. There have been several attempts to explain the residual absorption in the superconducting state assuming d-wave symmetry. In a d-wave superconductor, the DOS is finite at all finite

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

477

frequencies, which is the necessary condition to account for non-vanishing absorption. However, detailed calculations for a d-wave superconductor show that the experimentally observed residual conductivity in the low-frequency region is still extremely large for any realistic choice of the scattering rate (Carbotte et al. 1995, Quinlan et al. 1996). Indeed, free-carrier absorption is prohibited in a clean metal by momentum conservation. Absorption is possible if impurities are introduced into the problem. But impurities are detrimental to the transition temperature of a d-wave superconductor (Rojo and Levin 1993, Radtke et al. 1993). Also, it is known that even minor concentrations of impurities lead to changes of the temperature dependences of the penetration depth )~,b(T) from linear in clean crystals to quadratic is disordered materials (Bonn et al. 1993). In this context, it is important to emphasize that the residual absorption is found in samples with the highest Tc for the Y123 series that exhibit a linear law for the penetration depth (Zhang et al. 1994, Basov et al. 1995a). In summary, the simple d-wave picture is capable of accounting for many important trends in the ab-plane response but not the residual conductivity in the low-frequency region. Recent calculations with resonant scattering from impurities, to be discussed below in the section on impurities, do offer a possible mechanism that accounts for the finite low-frequency conductivity (Lee 1993 and Branch and Carbotte 1998). It is interesting to compare O'dc(T ) and ol (co) in the limit of zero temperature for superconducting crystals. The latter quantity can be determined relatively easily from the infrared measurements. But in order to obtain C~ac(T---+0), superconductivity has to be suppressed. This has become possible due to recent progress in pulsed magnetic-field techniques which are now capable of achieving fields up to 60 T. Such fields are sufficient to suppress superconductivity in materials with Tc around 3 0 K (Boebinger et al. 1996) including Lal.85Sr0.15CuO4. Experiments by G. Boebinger and collaborators revealed a cross-over of p(T) at T < Tc to a non-metallic dependence: ado(T) = 1 / P d c ( T ) decreases with decreasing temperature. This suggests that the material has an unusual ground state at T --+ 0. Measurements of the infrared conductivity for the same material (Somal et al. 1996) clearly show that cr1(co) increases with decreasing frequency. This latter behavior is consistent with a metallic ground state. We note that usual transport theories result in a similar T and co dependence of the conductivity (Pines and Nozi~res 1966). As discussed in sect. 4.3 this T and co behavior of the conductivity is similar in cuprates at T > Tc. A comparison of the results of Somal et al. (1996) with the data obtained by Boebinger shows that this correspondence is no longer observed at T < T~ at least in La214. If this result is confirmed by measurements on the same sample, it will be indicative of completely novel transport phenomena in high-To cuprates. We now turn to the discussion of superconducting-state properties for the case where the electric field and the induced currents are along the c-axis. The charge dynamics along the c-axis are incoherent, which in principle simplifies the search for the superconducting gap. At the time of writing, c-axis results in the superconducting state are available for Y123 (Bauer 1990, Homes et al. 1993b, Schfitzmann et al. 1994, Hauffet al. 1996, Lobo et al. 1996, Miinzel et al. 1996, Homes et al. 1995a) for Y124 (Basov et al. 1994a, Tajima et al. 1998), Bi~Sr2CaCu208 (Tajima et al. 1993), for La214 (Tamasaku et al. 1992, Gerrits

478

D.N. BASOVand T. TIMUSK

et al. 1994), Uchida et al. 1996, Basov et al. 1995b, Henn et al. 1996, Schfitzmann et al. 1997, Birmingham et al. 1996), Pb2 Sr2ReCa2 Cu3 08 (Reedyk et al. 1994), T12Ba2CuO6 + 6 (Schfitzmann et al. 1997), and La2euO4+6 (Quijada et al. 1995). In all these materials, the conductivity remains finite down to lowest frequencies and there is no evidence for the superconducting energy gap. Thus, the c-axis experiments (figs. 11, 12) reinforce the conclusion of the gapless response inferred from the in-plane studies. There have been attempts to account for gapless c-axis conductivity with models where the order parameter is anisotropic. For the optimally doped compounds these attempts have had some success. Homes et al. (1993b) concluded that the superconducting-state conductivity of YBa2Cu306.95 is consistent with the calculations performed for a p-wave superconductor (Hirschfeld et al. 1989). The principal feature of the data, which is in accord with a p-wave model, is that the conductivity varies linearly with frequency at T << T~. Uchida (1997) also observed a linear variation of c~1(0)) in the optimally doped Lal.85 Sr0.15CuO4 crystal. They fitted the results with a model for a d-wave superconductor (Graf et al. 1995a,b). However neither d- nor p-wave models account for the conductivity in the underdoped Y123 samples. Indeed, in these materials the spectra of o1(o)) are dominated by the pseudogap which occurs in the normal state at T < T* (fig. 11). An interesting feature of the conductivity spectra obtained for underdoped crystals (both Y123 and La214) is the similarity of the normal-state results obtained at T~ < T < T* and the superconducting-state data. The same conclusion follows from the discussion of the in-plane conductivity of underdoped crystals shown in fig. 7. It is often asserted that the similarity between superconducting- and normal-state data for underdoped cuprates is related to the fact that the pseudogap is a precursor of the superconducting gap. This "precursor hypothesis" is inconsistent with the c-axis conductivity of the Y123, Y124, and La214 series. At least in these materials, the formation of the pseudogap is arrested at T ~ T~ (Basov et al. 1995b, Homes et al. 1995a). This conclusion follows from the analysis of the destination of the spectral weight at T < T* and Tc that can be determined from the imaginary part of the conductivity. When the spectral weight is transferred below the low-0) boundary of the measurements, the imaginary part of the conductivity is enhanced. The closest analogy of this process is the formation of the superconducting delta function at 0) = 0 which leads to the increase of cr2(0)) depicted in fig. 15. On the contrary, when spectral weight is transferred to higher energies, 02(0)) is suppressed and can even become negative. The development of a pseudogap in the c-axis response is always associated with the latter process. On the other hand, suppression of al (0)) in the superconducting state is associated with the formation of a superconducting condensate, and the spectral weight missing from a~(0)) at finite ~equencies is recovered under the delta function at 0)=0. Also, the two processes have markedly different energy scales. While the characteristic scale related to the pseudogap in the Y123 and Y124 series is about 300cm -1, the superconducting condensate is accumulated from energies extending up to at least 1000cm -1 (Basov et al. 1995b). Perhaps the most intriguing feature of the c-axis response in the superconducting state is the development of a "plasma edge" (Bonn et al. 1987) in the far infrared. The effect is most clearly observed in the La214 system (fig. 16). In these strongly

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW meV 25

50 '

75

100

125

150

'

'

YBa'Cu306'952

175

4000 E II a T=T

~ ( ~ 1 (c°) E .

0 10000

2000

.

.

.

.

.

T=10K

479

Fig. 15. Infrared probe of the penetration depth in superconductors. The top panel shows the frequency dependence of the real part of the conductivity obtained for the polarization of the incident radiation along the a-axis of a YBa2Cu306.9s single crystal. Thick cmwe: T ~ T~; thin curve: T << T~. Upon transition to superconducting state, part of the spectral weight (shadowed area) is transferred to a delta function at co = 0. This is also indicated by a characteristic I/co dependence of the imaginary part of the conductivity (middle panel). That the 1/co law for ~r(co) is indeed followed in the superconducting state is illustrated in × 0"2(0) ) the plot of ),(co) = (bottom panel). In the limit of zero frequency, ,1(co) extrapolates to the penetration depth which can alternatively be determined from the stun rule for the spectra of crl(co) (eq. 14). Experimental results are from Basov et al. (1995a).

c/,~/~

,sum rule

)~(co) A I

I

I

1

T

I

I

200

400

600

800

1000

1200

1400

W a v e n u m b e r [cml "1

anisotropic materials, the reflectance spectra taken in the normal state are "insulator-like": one finds several strong phonon modes at co > 200 cm : and a flat, frequency-independent background in the far infrared where R ~ 0.5 (Tamasaku et al. 1992, Basov et al. 1995b, van der Marel and Kim 1995, Gerrits et al. 1995, Birmingham et al. 1996, Uchida et al. 1996). There is very little temperature dependence in the spectra as T decreases from 300 K down to T ~ T~. However, as soon as these samples undergo the superconducting transition the reflectance immediately changes its character: R nearly reaches unity at the lowest frequencies and then abruptly drops down to R ~ 0.2. At higher energies, the reflectance slowly recovers to the magnitude seen in the normal state. This behavior resembles the plasma resonance that occurs in ordinary metals at co ~ cop. Two different approaches have been used to explain the sudden appearance of this feature in the superconducting reflectance. As originally pointed out by Bonn et al. (1987) the sharp feature originates from a zero crossing of the real part of the dielectric function resulting from an interplay between the positive phonon contribution and the large negative contribution of the superconducting condensate. In the normal state the c-axis plasmon is strongly damped, but at T < T~ the scattering rate suddenly drops and the plasma resonance becomes observable (van der Marel and Kim 1995). An alternative scenario is to interpret this feature as the result of Josephson coupling between the CuO2 layers (Tamasaku et al. 1992, Basov et al. 1994a). Employing a theory developed in the early 1970s for the electrodynamics of an array of Josephson-

480

D.N. BASOV and T. TIMUSK

1.0

~--~

0.5 -

!a-

x=0.30

_ 2 -

............ aUove c f

v \

~=o.ls

=

"

-

-

r=8~:

x=0.10 x=0.12

Frequency (cm-1) 1.0

0.8

°= o.6

0.4

0.2

0.0

T

~

c

=34K

'

~

50

100 Wavenumbers, [cm] "1

'

500

'

Fig. 16. Top panel: spectra of the c-axis reflectivity measured above (T ~ 40 K, dashed curves) and below T~ (T = 8 K, solid curves) for various compositions of La2 xSr, CuO4 single crystals between 0.10 and 0.30 (from Uchida et al. 1996). Bottom panel: spectra of the c-axis reflectivity for a single crystal of La2_~Sr, CuO 4 with T~ = 3 4 K measured at 8K, 17K, 20K, 22K, 25 K, 27 K, 32 K and 35 K (from Basov et al. 1995b).

coupled junctions (Lawrence and Doniach 1971, Bulaevskii 1973), it is possible to reproduce the frequency dependence of the observed reflectance. An obvious difference of cuprates from artificial Josephson contacts is that the "normal" barriers inside the unit cell of the crystal have thicknesses of the order of interatomic distances. This shifts the plasma edge from GHz frequencies, which are typical for artificial junctions of conventional superconductors, to the THz range (20-200cm -I) in cuprates. One also observes that the frequency position of the plasma resonances scales with doping in a series of La2-xSrxCuO4 crystals from 20cm / in underdoped samples up to 120 cm-1 in overdoped materials. The consequences of Josephson currents for the c-axis penetration depth will be discussed in the next subsection. Concluding this subsection, we stress that there is no experimental evidence for a clear superconducting energy gap in any of the cuprates. Residual absorption is universally found in high-T~ superconductors. This behavior is different from the response of other

1NFRAREDPROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

481

novel superconductors such as K3C60 and R b 3 C 6 0 (Degiorgi et al. 1994, Koller et al. 1996) or borocarbides (Bommeli et al. 1997) which all show strong evidence in favor of a BCS gap in the infrared conductivity. Theoretically, the response of cuprates is better understood within a d-wave model, while, in our opinion, we still have a long way to go before the calculations yield a self-consistent explanation of infrared and microwave properties. 5.2. Anisotropic penetration depth in cuprates Regardless of the microscopic details of the superconducting mechanism, the transition into the superconducting state is associated with the formation of a delta function in the real part of the conductivity at co = 0. From the point of view of the electrodynamic response, this implies that the spectra of ol (co) must be suppressed at finite frequencies, with the spectral weight being transferred to the delta function. "The missing spectral weight" is identical to 1/3`2, according to the superconducting-state sum rule (Tinkham and Ferrell 1959). Alternatively, the magnitude of 3` can be inferred from the imaginary part of the conductivity since a delta function al (0) with the spectral weight 1/)~2 implies by KK that 02(0)) ec 1/(3`2 × co). Note that the penetration depth 1/3,2 = nse2/m is not equivalent to the London penetration depth defined as 1/2~ = ne2/m. The difference arises from the fact that the density of superconducting carriers ns is usually smaller than the total carrier density n leading to 3`L < 3`. The two quantities coincide only in the clean limit when the width of the Drude peak in al(co) is much smaller than the energy gap. In this case it is usually concluded that "all free carriers condense." Figure 15 illustrates the two approaches towards the determination of the penetration depth from infrared where data for the a-axis conductivity of the YBa2Cu306.95 compound (Basov et al. 1995b) are shown. The real part of the conductivity al(co) is suppressed at T < ~. That the missing spectral weight (shadowed in the diagram) is transferred to a delta function at co = 0 is indicated by the frequency dependence of 02(0)) in the superconducting state. It acquires the characteristic 1/co dependence which signals that the spectral weight is recovered in the delta function. To show the presence of a 1/co term it is often instructive to plot the product of 1 / ~ , which in the zero-frequency limit yields the value of)~. Thus, the penetration depth or the superfluid density can be obtained from: (i) the sum rule for the optical conductivity, c - cops 2 - 120 3`2 yc f0 °~ do) [ol(co, T ~ Tc) - O'l(co, T ---+ 0 ) ] ,

(14)

and/or (ii) from examining the prefactor in front of the l/co term in the imaginary conductivity: c

3`2

_

= co ×

2(co, r

-,

0).

(15)

482

D.N. BASOV and T. TIMUSK YBa2Cu408

t 6000

4000 EIIb

E o "T E .¢:

o

EIla

2000

0 - 6000

0 YBa2Cu306,95

4000

==

EIIb

g x

-

2000

.>_ 0

0

o E

4000

E[Ib

6000

YBa2Cu307. x

4000

4000

g

2000 Ella i

I 1000 0

1000

0 2000

Wavenumbers (cm "1)

Fig. 17. Left-hand panels: Frequency-dependent penetration depth ,t, - c/[~o)(e))] for a Y124 sample (upper panel), and two different Y123 samples (middle and bottom panels). Right-hand panels: the real part of the complex conductivity . All curves shown are at 10K. The left-hand panels reveal clear anisotropy of the penetTation depth which is stronger for the double-chained Y124 material. From Basov et al. (1995a).

In our experience, in the case of the ab-plane response the sum-rule result for the penetration depth (eq. 14) often yields a larger value than the estimate from o2((o). However, since the difference usually does not exceed 10% it may be premature to put too much trust in this result until the experimental accuracy is improved. A unique advantage of infrared spectroscopy for studies of the penetration depth in cuprates is that all diagonal components of the penetration-depth tensor can be obtained from model-independent analysis of infrared data for single-crystal samples. The analysis using eqs. (14) and (15) can be applied to the conductivity probed with polarized light for the polarization of E-vectors both along the CuO2 planes and along the c-axis. In the YBCO series, there is an additional anisotropy associated with the Cu-O chains. Infrared experiments carried out with incident radiation polarized along the chain direction (b-axis) and perpendicular to the chains (a-axis) made it possible to extract both )~ and )~b (Basov et al. 1995a). The frequency dependence of ,~(co) = c/v/(a2(co) × co) for untwinned Y123 and Y124 crystals with polarizations of incident radiation along a- and b-axes is shown in fig. 1Z In the limit of zero frequency, A(co) extrapolates to the penetration depth of a

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

483

superconductor. An interesting result is that the penetration depth is strongly suppressed when it is probed along the direction of the Cu-O chains (E tl b). in the doublechained Y124 crystal the anisotropy is larger. These observations are consistent with the notion that the chains are superconducting in YBCO compounds (Basov et al. 1995a). In high-quality YBa2Cu306.9s crystals the anisotropy of the penetration depth is approximately the same as that of the dc resistivity or of the normal-state plasma frequency: )~/Xb 2 2 ~ rhoa/pb ~ OOpb(Dpa 2 2 ~ 2.3. While the absolute values of ~a and )~b are different, the temperature dependence of these quantities is identical based on microwave measurements (Zhang et al. 1994). The frequency dependence of c~1(~o) is also very similar for the two polarizations. This indicates that the electrodynamic response of YBa2Cu30695 is identical along the a- and b-directions, except for the additional spectral weight along the a-axis. That the 1D Cu-O chains contribute to the superfluid density is a robust experimental observation first found with IR spectroscopy (Basov et al. 1995a) and later confirmed with muon-spin relaxation (~SR) spectroscopy (Tallon et al. 1995) and Josephsontunneling measurements (Sun et al. 1995). With scanning tunneling spectroscopy, it was possible to detect a gap-like feature for the Cu O chains similar in magnitude to that for the CuO2 planes (Edwards et al. 1995). The interpretation of these results indicating superconductivity in the Cu-O chains is difficult. Most theoretical models start from the notion that the pairing mechanism is an exclusive property of the CuO2 planes, whereas other elements of the unit cell are regarded as passive placeholders. Within this approach, one attributes the formation of the superconducting condensate in the chains to a proximity effect between "intrinsically superconducting" CuO2 planes and "intrinsically normal" Cu-O chains (Kresin and Wolf 1992, Xiang and Wheatley 1995). The major problem with this interpretation is that proximity-based models cannot account for the identical temperature dependences of 3~ and Xt, found with microwave spectroscopy (Zhang et al. 1994). Indeed, one expects that proximity coupling induces a gap in the normal layer that has suppressed magnitude (Kresin and Wolf 1992, Atkinson and Carbotte 1995). This will inevitably lead to a difference in the T-dependence of the penetration depth (Atkinson and Carbotte 1995) and to a difference in the frequency dependence of al (co) (Atkinson and Carbotte 1997). Both of these expectations, based on the proximity-coupling model, are inconsistent with experiments. The failure of the proximity-based scenario to account for experimental observations may be suggestive that the Cu-O chains are also intrinsically superconducting. In fig. 18, we summarize the experimental results for the in-plane penetration depth obtained with infrared spectroscopy for a variety of cuprates. There are two principal trends revealed by this plot: (i) In many materials, Tc scales with the superfluid density. This was first discovered by Uemura et al. (1989, 1991) based on the analysis of the ~SR data. (ii) In YBCO materials containing Cu-O chains, only the a-axis superfluid density correlates with To. The anisotropy of the penetration depth may account for the "plateau regions" in the original Uemura plot observed for samples close to optimal doping. In disordered crystals, the linear dependence between the superfluid density

484

D.N. BASOV and T. T1MUSK

___=- ".?...q "--2~ ......

100.0

[] 1 3 4 D2212 z~ ps2

10.0

,<~

1.0

"~

~5 123

0

123

0

o 123

""

6.7 6.85

,,,,,,,,I

, ,,1,..I

0.0

0.1

I

/I

TI2201

• ~

, ,,,,H,l

t.0 10.0 O'FIR ( O e r n ) -1

•SR relaxation rate, gs

100

"" ....

o 6.95

, ,,l..I

0.1

""~'"-.. ~'~"--..

/

/

/

4 I

-1 8 I

I

, ,

100.0

I

/ "universal" plot

¥12a°6.gs

#

==l

¢ Y124

I~

Y124066

RbC60 KC60 0

La214 /

"/t" NdCeCuO I

lx108

I

I

2xt08

2, cm

3x108

I 4x108

-2

COps

Fig. 18. Upper panel: Correlation between the values of the penetration depth Zc and the c-axis conductivity of different high-~ superconductors. Lines are calculations for the case of Josephson coupling between the CuO 2 planes assuming 2A = 128 cm-I (short dashes) and 2A = 320cm -1 (long dashes). From Basov et al. (1994a). Lower panel: in-plane superftuid density (%s in a variety of high-T~ superconductors. For the purpose of comparison with analogous plots based on the p~SRdata, units for the top axis are given in p~SRrelaxation rate. In many cuprates the critical temperature appears to be linearly dependent on ns/m* ("universal" Uemura plot). The b-axis superfluid density in the Y123 and Y124 sex'ies (diamonds) as welt as the results for the electron-doped Nd2_yCevCuO4 show remarkable deviation from this plot. Data: Y123, Y124, Basov et al. (1995a); T12201, Puchkov et al. (1995); Bi2212, Quijada et al. (1994); BaKBiO, Puchkov et al. (1996d); La214, Startseva et al. (1999a); RbCa0, KC60, Degiorgi et al. (1994). and To is v i o l a t e d ( N a c h u m i et al. 1996, Bernhard et al. 1996) as will be discussed in subsection 5.3. A n o t h e r notable deviation f r o m the universal p l o t is found in electrondoped Nd2_yCeyCuO4 ( H o m e s et al. 1997). The superfluid density in this material significantly exceeds that o f other superconductors w i t h T~ ~ 20 K. It should be noted that within the B C S theory one does not expect to observe any explicit correlation b e t w e e n To and the superfluid density in a c l e a n - l i m i t superconductor. T h e linear d e p e n d e n c e b e t w e e n the transition temperature and ns/m* is e x p e c t e d i f

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

485

superconductivity is viewed as Bose Einstein condensation. The analysis of the absolute values of the penetration depth allows one to estimate the number of carriers per area of the coherence length squared, ~b, on the conducting ab-plane (Uemura 1996). Both cuprates and organic superconductors have only several superconducting pairs per ~2, which puts these systems close to the limit of Bose-Einstein condensation (1 pair per 1/~2 rather than to the limit of BCS condensation where one finds more than 10 000 carriers per ~2). Emery and Kivelson (1995b) emphasized that in superconductors with low superfluid density (such as cuprates and organic materials), the role of phase fluctuations is dramatically enhanced. They argued that in the above materials the magnitude of the superfluid density controls the temperature of long-range phase order and thus determines Tc. This accounts for the striking deviations of the transition temperature from typical mean-field behavior. The c-axis penetration depth is strongly enhanced in all high-Tc superconductors compared to 2,b (fig. 18). Typical values of ;.c range from 100 ~tm in the most anisotropic Bi2212 crystals (Motohashi et al. 2000, J.R. Cooper et al. 1990) to about 1 ~tm in the most isotropic YBa2Cu30695 and Y124 crystals (Homes et al. 1993b, Basov et al. 1994a, Schfitzmann et al. 1994). The magnitude of Xc shows a systematic correlation with the dc conductivity along the c-axis, ac (Basov et al. 1994a). It was emphasized that the proportionality ;~c2 o( A x ~. is expected for an array of Josephson junctions (Basov et al. 1994a). Thus, Josephson electrodynamics successfully describes the main features of the c-axis penetration depth in a broad variety of cuprates. Recently, Tajima et al. (2001) have used the c-axis Josephson plasma resonance in Nd-doped Lal.85-yNdySr0. lsCuO4 to show that stripe order in this material suppresses the interlayer phase coherence. The anomalous c-axis properties of high-Tc superconductors and the strong evidence in favor of Josephson coupling between the CuO2 layers give experimental support for the "confinement hypothesis" of high-Tc superconductivity (Chakravarty and Anderson 1994). Within this hypothesis, the hopping of single electrons between C u Q planes is inhibited in the normal state by spin-charge separation. However, tunneling of pairs is allowed at T < T~, which leads to a gain of kinetic energy as Cooper pairs are formed and the superconductivity sets in. This energy is the Josephson coupling energy, and it is proportional to the square of the plasma frequency of condensate probed for the polarization where the E-vector is along the c-axis. Thus, if interlayer tunneling of Cooper pairs is the driving force for superconductivity, one expects to find a correlation between Tc and 1/;~ (Anderson 1995). A brief examination of fig. 18 suggests that, in general, such a correlation is not followed in cuprates. For example, the Tc of the most anisotropic Bi2Sr2CaCu208 crystals is nearly the same as in the most isotropic Y123 and Y124 compounds. An accurate determination of the superfluid density in Bi2212 was done by Tsui et al. (1996) using microwave techniques; they found a plasma frequency of about 5 cm 1. Within the interlayer-tunneling models these estimates are inconsistent with the T~ value of 90 K. However, it should be noted that the correlation between Tc and the superfluid density expected within the confinement hypothesis may be complicated in Y123 and Bi2212 systems where there are two CuO2 layers per unit cell.

486

D.N. BASOVand T. TIMUSK

Careful examination of the spectral-weight balance at T < T~ using eq. (14) and eq. (15) suggest that in underdoped cuprates the superfluid density (the area confined in the delta peak at co = 0) exceeds the weight missing from the real part of the conductivity Basov et al. (1999). This result can be interpreted in terms of kinetic energy change AK (Hirsch 1992, Anderson 1995, Chakravarty 1998). However the absolute values of AK are significantly smaller than the condensation energy, as first pointed out by van der Marel (1996) and later discussed in several other publications (Schiitzmann et al. 1997), Moler et al. 1998, Tsvetkov et al. 1998, Kirtley et al. 1998). 5.3. Temperature dependence of the penetration depth Using infrared spectroscopy, it is virtually impossible to obtain optical constants of a material at more than a few different temperatures. This is an obvious disadvantage of infrared techniques compared to microwave methods which allow one to study the temperature dependence of the penetration depth and of the surface resistance in great detail (Bonn and Hardy 1996). However, it is usually difficult to determine the absolute values' of the )~-tensor with microwaves. This drawback of microwave methods is easily overcome with infrared techniques which yield all the diagonal components of)~ without the need to take into consideration any geometrical factors. A combination of infrared methods providing the absolute value of ,~a(T -+ 0), ).b(T ---+ 0), and )Lc.(T --+ 0) with microwaves, giving a detailed and accurate temperature dependence A)~ = )~(T) - )~(0), is becoming an indispensable tool in studying the penetration depth in high-To superconductors. Figure 19 shows the temperature dependence of A~. for a-, b- and c-polarizations measured from the frequency change of a superconducting loop-gap resonator (J~ ~ 900 MHz) (Hardy et al. 1996). IR-determined values of ,~0 are shown in table 2. The data set for overdoped YBazCu306.99 is an exception where infrared measurements have not yet been done. At all doping regimes, both a and b components of the )~-tensor reveal linear behavior which was first observed in twinned YBa2Cu306.95 samples (Hardy et al. 1993). The linear variation of the penetration depth is different from what is expected for a BCS superconductor with an isotropic gap. An isotropic gap leads to an exponential dependence of,~(T). While the thin-film specimens studied prior to 1992 tended to show a T 2 law for )~(T), a series of more recent studies performed on high-quality single crystals show a T-linear law for ~.(T) (Ma 1995, Jacobs et al. t995). A consensus is emerging in the field that the earlier observations of a T 2 dependence were dominated by impurity effects, whereas intrinsic behavior of the penetration depth can be found only in clean samples, and that variation is linear with T. This problem was studied in detail by the group at the University of British Columbia. Starting from a high-purity YBa2Cu306.95 single crystal with A,~,h ec T, Bonn et al. (1994) deliberately added Zn and Ni ions which are believed to substitute for Cu in the CuO2 planes. They observed a cross-over to a A,~a/~(T) ec T 2 law in the crystal, containing as little as 0.31% Zn. Recent thin-film samples also show a linear penetration depth at low temperatures. This behavior was first observed by Gao et al. (1993b) in their microwave experiments

I N F R A R E D P R O P E R T I E S OF H T S C : E X P E R I M E N T A L O V E R V I E W 1.0

487

(a)

0.8

o 0.6

0

X=6.6

• 0

oa

~1

0.4

•b ,¢

O

¢

% 0.2

0.0

I

40.0

20.0

0.0

"r(N 1.0

o

• •

o

"

0.8

• •

:

•

0 0 0

• 0

0.6

•

X~.95

0 0

.....

• o

o

•

oa 0.4

• O0 e

.~

•

•b *0

0.2

0.0 0.0

i

i

I

20.0

40.0

60.0

80.0

1.0

(e) •

0.8

o

o

•

o o

0 0.6

X=6.99

•

• 0

•

\

0 •

o

oa

0.¢

•

•b *¢

°o

0.2

0"00.0

i

i

i

20.0

40.0

60.0

80.0

Fig. 19. ~,2(0)/.,3,2(T) versus T for a, b, a n d c directions in Y B a z C u 3 0 x single crystais. The absolute values o f ~(0) are given in table 2. F r o m H a r d y et al. (1996).

488

D.N. BASOVand T. TIMUSK

Table 2 Values of )~(0)obtainedwith IR spectroscopy(Basovet al. 1995a, 1996) used to generate,~2(0)/;t2(T) in fig. 19 based on microwavemeasurements x

~ (K~

£(0) (X)

Xb(0)(A/

Xc(0)(X)

6.60 6.95 6.99

59 93.2 89

2100 1600 1600~

1600 1030 800"

65000 11000 11000 h

'1 Value estimated based on p~SRresults for ,~a/3,~(Tallonet al. 1995). h Value measured directly using microwavetechnique. conducted at 10GHz. Dahne et al. (1995) and de Vaulchier et al. (1996) used backwardwave oscillators operating at frequencies from 3 c m l up to 15cm -1 to examine the temperature dependence of the penetration depth in thin-film samples of YBa2Cu30695. They also observed a linear behavior of ,~ab(T). These recent observations are clearly consistent with the notion that the T 2 variation found in the earlier samples was of extrinsic origin and only high-purity crystals and films of YBCO show a linear penetration depth variation with temperature. A penetration depth which varies linearly with temperature is expected for a superconductor with d-wave symmetry of the order parameter (Prohammer and Carbotte 1991, Hirschfeld et al. 1994). There are also a number of alternative explanations. For instance, a linear term in ,~(T) can be caused by the presence of normal 2D layers sandwiched between the CuO2 planes (Klemm and Liu 1995). Another possibility is that a d-wave response can be mimicked by the complicated band structure of the YBCO system involving both plane and chain bands (Combescot and Leyronas 1995). It was also pointed out that Josephson coupling in a bilayer system can, in principle, lead to vanishing of the gap function at certain momenta and thus bilayered materials may imitate properties of d-wave superconductors without invoking a d-wave state (Gauzzi and Bok 1996). Finally, a number of theories suggest that the linear )~(T) can originate from fluctuations of the phase of the order parameter (Radtke et al. 1993, Roddick and Stroud 1995, Coffey 1995, Emery and Kivelson 1995a). Further work is required to distinguish between the proposed models. There has been one suggestion, based on microwave measurements, that the intrinsic behavior of the penetration depth in Y123 crystals can be different from linear. Srikanth et al. (1997) reported a non-monotonic behavior of,~(T) in single crystals grown in BZO crucibles. Such crystals are claimed to have superior purity compared to the previous generation of Y123 samples grown in yttria-stabilized zirconia (YSZ) crucibles. So far, the experimental results are limited to twinned samples at optimal doping. However, a recent result on untwinned samples, also grown in BZO crucibles, fails to show this unusual behavior (Kamal et al. 1998). Several groups have reported measurements of the temperature dependence of the c-axis penetration depth in YBCO (Basov et al. 1994a, Mao et al. 1995, Hardy et al. 1996, Lobo et al. 1996, Homes et al. 1997), Bi2Sr2CaCu208 (Jacobs et al. 1995) and La214

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

489

(Shibauchi et al. 1994, Basov et al. 1995b, Henn et al. 1996). The present experimental situation is controversial. In part, the variation in the data from different groups can be attributed to anomalous sensitivity of the c-axis properties of cuprates, even to minor variations of doping.

5.4.

Impurity effects

A desire to establish the symmetry of the superconducting order parameter was the driving force of a large number of experimental programs. Among them were numerous attempts to distinguish between s- and d-pairing through studies of the effect of impurities on the transition temperature, the complex conductivity and the T-dependence of the penetration depth. Non-magnetic impurities have little effect on the superconductivity in elemental metals (Anderson 1958). On the contrary, anisotropic superconductors are expected to be anomalously sensitive to impurities or disorder, and impurities lead to a smearing of the gap anisotropy towards some average value. In particular, the average of the d-wave gap is equal to zero and, therefore, in a d-wave system, superconductivity is easily suppressed even with minute concentrations of impurities. In qualitative agreement with the picture of anisotropic superconductivity, the value of Tc in cuprates is diminished when thin films or single-crystalline samples are doped with non-magnetic Zn, or disordered by irradiation with low-energy ions (Valles et al. 1989, Sun et al. 1994, Tolpygo et al. 1996, Moffat 1997, Fukuzumi et al. 1996, Yoshizaki et al. 1996). Another theoretical prediction for d-wave superconductors is that impurities cause severe changes in the intragap density of states (Borkowski and Hirschfeld 1994, Fehrenbacher and Norman 1994, Preosti et al. 1994a,b, Palumbo and Graf 1996). Therefore, one expects to find dramatic effects of impurities on the temperature dependence of the penetration depth and on the frequency dependence of the complex conductivity, which both probe the low-energy excitations in a superconductor. Early experimental studies of the infrared properties of deliberately disordered films of Y123 include the paper by Lippert et al. (1991) which was followed by the work of Sumner, Kim and Lemberger (1993) on Ni-doped Y123 films, and experiments on disordered Bi2212 thin films (Mandrus et al. 1993). Prior to 1992/93, when the above experiments were performed, it was widely believed that the high-T~ superconductors were s-wave. In the s-wave dirty limit an observation of the superconducting energy gap should be possible (Timusk and Tanner 1989, Kamarfis et al. 1990, Tanner and Timusk 1992). Nevertheless, the structure at 2A that can be attributed to a conventional BCS gap did not appear in transmission or reflectance spectra even when impurity scattering was significantly enhanced by disorder. Instead, experiments showed that in disordered specimens, the distinction between the spectra in the normal and superconducting states became less pronounced (Ulm et al. 1995, Mandrus et al. 1993, Kim et al. 1994). The materials, in effect, acquired a strong "normal component" in the superconducting state. Basov et al. (1994b) reported measurements of the ab-plane reflectance of a highquality YBa2Cu3069s single crystal which they irradiated with low-energy He ions

490

D.N. BASOV and T. TIMUSK

eV 0.00 0.05 0.10 0.I5

0.05 0.10 0.15

•

i

F ....

' .............

nonirradiated T==93.5 K

[0000

o

>Z g

i

i

i

3 . . . . . . . . . .

I,:~

''"J

2xl.0" H e / e r a ' ~ T,=85 K

5000

0

toooo

,,

--

.o/cm' T,=80 K ~

•

T,=70 K

5000

o I-, 0

, "-,-,

.

.

.

.

500

.

.

.

.

.

.

.

.

.

,

....

t000 0 500 1000 Wave n u m b e r s ( c m - ' )

1 1500

Fig. 20. Frequency dependence of the conductivity of He+-irradiated single crystals of YBa2Cu3069> Upper curves in all panels were measured at temperatures slightly above T~, while the lower curves were obtained at 10 K. The dashed curves show the fit of the narrow normal component in the superconducting conductivity to the Drude formula. After Basov et al. (1994b).

(fig. 20). Based on these measurements, they obtained the far-infrared conductivity which was analyzed within a phenomenologieal two-fluid model: i(2~c) 2 co2. r(co)/4= O(co, T) - ~ COA2(T + 1 + icor(co)

(16)

In eq. (16), the first term is imaginary and stands for the response of the superftuid with the spectral weight given by 1/,t 2. The second term describes the residual absorption of the normal component with the simple Drude formula with the plasma frequency COp,,and 2 + l/t12. the scattering rate 1/T. The sum rule fixes the total plasma frequency CO~D= COp, At frequencies below 150 cm q , the co-dependence of 1/T can be neglected and, therefore, eq. (16) is adequate for the description of the low-energy results. It was found that ion damage reduced the amount of superfluid density and simultaneously enhanced the plasma frequency COp, associated with the normal component in the superconductingstate conductivity (fig. 20). The reduction of superfluid density in ion-damaged samples is consistent with the notion of pair breaking due to non-magnetic impurities. Basov et al. also concluded that irradiation has introduced sufficient impurity scattering so that an s-wave gap would be easily resolved in the spectra of ol (co). However, there are no features in the spectra that could be attributed to such a gap. This is independent evidence

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

491

that Y123 is not an s-wave superconductor. Resonant scattering by impurities in a d-wave superconductor produces a finite density of states at the Fermi surface of magnitude 1/T. This, in effect, cancels the T factor in the Drude formula giving rise to a universal conductivity which is independent of impurity content (Lee 1993). Recent calculations of the optical conductivity in the presence of resonant scattering (Branch and Carbotte 1998) suggest that this mechanism may well account for the residual optical conductivity at low frequency in the high-temperature superconductors. Carbotte et al. (1995) analyzed the complex conductivity in the presence of disorder in the weak Born-scattering limit assuming different symmetries of the order parameter. Their main result is that the data presented in fig. 20 are not consistent with an s-wave gap but can be qualitatively accounted for within a d-wave model. In particular, disordered d-wave superconductors are expected to reveal an enhancement of the spectral weight of the "normal component" in the superconducting state response. The frequency dependence of the conductivity at T << Tc and its evolution with disorder are remarkably similar in the experimental data by Basov et al. (1994b) and in theoretical calculations by Jiang et al. (1996). Palumbo and Graf (1996) considered an anisotropic layered superconductor which can be described as an array of superconductor-insulator ( S - I - S ) junctions. They calculated the superconducting conductivity in the presence of impurities for three different types of order parameters: anisotropic s-wave (ASW) Aasw(0) = A0(1 + cos(40)/2), extended s-wave (ESW) AEsw(0) = A0(1 + 3 cos(40)/4), and d-wave (DW) ZlDw(0) = A0 cos(20). All three order parameters have nodes on the Fermi surface, but only AEsw(0) and ADw(0) change sign. In order to be qualitatively consistent with the data at T << T~, the change of sign of the order parameter is required. A similar conclusion was inferred from earlier theoretical work in the context of studies of the density of states in a disordered superconductor with anisotropic order parameter (Borkowski and Hirschfeld 1994, Fehrenbacher and Norman 1994, Preosti et al. 1994a,b). In the ASW case, disorder leads to the opening of a true gap in the superconducting DOS. On the contrary, in the DW system, the finite density of states develops in the limit co --+ 0. Thus, the absorption of an ASW superconductor decreases in the presence of impurities contrary to the situation realized in ESW and DW superconductors. Comparison with experiment suggests that an ASW scenario can be ruled out, while both ESW and DW scenarios qualitatively agree with the data. Disordered YBCO materials reveal additional effects connected with the presence of the one-dimensional C u - O chains. As pointed out in sect. 5.4, in the first available untwinned single crystals, the chains were disordered due to contaminations with Au or A1 coming from the crucibles used in the crystal growth. In this case, the response of the chains was similar to what one expects to find in disordered 1D conductors (Schtitzmann et al. 1992). Shibata et al. (1995) have studied the a- and b-axis conductivities of an untwinned Y123 crystal doped with Zn. Their results for the clean crystal are in accord with the data by Basov et al. (1995a): the only difference between ala(co) and al~(co) is in the magnitude of the conductivity, whereas the frequency dependence is the same. However, this similarity disappears in the crystal containing 3% Zn. The b-axis response

492

D.N. BASOVand T. TIMUSK

of the sample with Zn substitution shows a characteristic peak which was found in the conductivity of crystals with Ag or Au contaminations. Thus, it is conceivable that Zn occupies some of the Cu sites in the chains. Recently, Wang et al. (1998) reported a strong effect in the b-axis conductivity produced by Zn doping, corroborating the idea that some of the Zn impurities reside on the chains. Homes et al. (1999) reported on the conductivity of untwinned Y123 samples doped with up to 1% Ni. Their main result is that Ni induces dramatic changes in the conductivity along the chain b-direction, whereas the a-axis response is almost unchanged, even for the samples with 1% Ni. This is also in accord with the view that it is difficult to avoid distortions of the chains in the course of impurity doping. Disorder also leads to a suppression of the ab-plane superfluid density in YBCO materials (Basov et al. 1994b, Ulm et al. 1995, Bernhard et al. 1996, Nachumi et al. 1996, Panagopoulos et al. 1996). The linear law between the magnitude of Tc and superfluid density in clean crystals is not observed in disordered crystals. In disordered compounds, COpsis suppressed much faster than the corresponding reduction of To. Ulm et al. (1995) analyzed the degradation of )~ab(0) in Ni- and Zn-doped Y123 and concluded that the increase of the in-plane penetration as a function of impurity concentration is much more rapid than the d-wave model predicts. Based on the ~tSR results for Zn-doped Y123, Bernhard et al. suggested that suppression of the superfluid density as the result of Zn doping is inconsistent with s-wave pairing but may be suggestive of a d-wave order parameter. Several authors discussed the possibility that the addition of impurities leads to a highly inhomogeneous state in the cuprates on a length scale comparable to the coherence length (Xiang and Wheatley 1996, Nachumi et al. 1996, Franz et al. 1997). Even minor concentrations of impurities lead to a complete suppression of the anisotropy of the penetration depth in the ab-plane of the Y123 family. This effect has been observed for both Zn and Ni doping (Homes et al. 1999, Wang et al. 1998). Basov et al. (1998) have reported on the a-axis conductivity of YBa2Cu408 crystals with Zn substitution for Cu. They show that Zn leads to the formation of a bound state in the response of the CuO2 planes at CO~ 10 meV The bound state corresponds to a peak in the real part of the complex conductivity which develops at the expense of the Drude-like excitation at CO= 0. Peaks at finite frequency arc often observed in strongly disordered conductors at the borderline of localization (Allen et al. 1975, G6tze 1981, Gold et al. 1982, Ng et al. 1986, Jang et al. 1990). Many of the cuprates which are intrinsically or deliberately disordered also exhibit a peak at finite energy in the al (CO) spectra. The former group includes twinned crystals and thin films of YBa2Cu306.95 (Basov et al. 1994b, Kim et al. 1994). Among intrinsically disordered systems showing a peak in the conductivity, one finds Pb2Sr2ReCa2Cu308 (Reedyk 1992), T12Ba2CuO6 (Puchkov et al. 1995) or Bi2Sr2 xLaxCuO6. (Weber et al. 1997). If the frequency position of these peaks in a variety of cuprates is chosen to characterize the energy scale related to localization, then this scale Ec does not exceed 120cm -1. The results of Basov et al. (1998) for Zn-substitmed YBa2Cu408 demonstrate that the spectral weight from CO< Ec does not contribute to the superconducting condensate. This suggests a connection between the metal-insulator transition induced by impurities and

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

493

the degradation of Tc in disordered superconductors. Transport measurements by Valles et al. (1989), Sun et al. (1994), and Tolpygo et al. (1996) confirm the occurrence in cuprates of many of the universal trends of the superconductor-insulator transition previously seen in ultrathin films of elemental superconductors (for reviews see Liu and Goldman 1994, and Valles and Dynes 1990). So far, the studies of impurity effects (both transport and infrared) were primarily focused on underdoped and optimally doped cuprates. Much less information is available on the effect of impurities on the c-axis conductivity. Hauff et al. (1996) analyzed the c-axis response of YBa2Cu306.6 material with Zn substitution for Cu. The choice of this particular compound is determined by the fact that pure crystals of YBa2Cu306.6 exhibit a well-defined pseudogap in the c-axis conductivity (see fig. 11). Another characteristic feature of the c-axis response of these crystals is a broad resonance located at co ~ 400cm -1. Homes et al. (1995b) and Hauff et al. (1996) suggested that this resonance is a product of strong coupling of oxygen modes to the electronic system. A similar mode is found in the response of Pb2Sr2(Y/Ca)Cu308 (Reedyk et al. 1994) and of YBa2Cu408 (Basov et al. 1994a) but not in La214 (Basov et al. 1995b, Uchida et al. 1996). As first discussed by Hauff et al. (1996), Zn substitution reduces the strength of this resonance, while the low-frequency behavior of the conductivity in the pseudogap region (co < 300 c m -l) is nearly unchanged. Zn-doped YBa2Cu408 crystals also do not show this mode although, in the latter case, the authors observed considerable changes in the electronic conductivity as well (Basov et al. 1996, 1998).

6. Infrared properties of HTSC in a magnetic field

A static magnetic field enters a type-II superconductor in the form of an Abrikosov lattice of vortices, each vortex carrying a quantum of magnetic flux q~0. For static fields normal to the ab-plane, the simple triangular lattice of conventional superconductors is slightly distorted due to the anisotropic penetration depth as seen in YBCO by Keimer et al. (1994) using small-angle neutron scattering. This distortion can be explained simply in terms of anisotropy of the penetration depth (Walker and Timusk 1995). In the case of YBCO, this anisotropy is large due to the superconductivity of the chains. At high temperature, near To, the flux lattice melts to form a fluid of vortices. For magnetic fields normal to the c-axis, the currents that sustain the vortices must jump from plane to plane. As we have seen in the section on c-axis properties, singleparticle transport in the c-direction is largely blocked in the normal state, but pair tunneling is possible through Josephson coupling of the planes in the superconducting state (Bulaevskii and Clem 1991). Optical experiments have been done in both magnetic field geometries. With the field normal to the planes, thin-film transmission experiments probe the influence of the static magnetic field on the ab-plane transport. The conductivity can be obtained from the power transmission by Kramers Kronig transformations or amplitude transmission by terahertz spectroscopy which yields the real and imaginary parts of the transmission

494

D.N. BASOV and T. TIMUSK

amplitude directly. To obtain the maximum information from the experiments, circularly polarized light is used to measure circular dichroism (Linh et al. 1996) or the measurements are carried out with the sample between crossed polarizers - a geometry that yields the off-diagonal component, axy, of the Hall conductivity (Spielman et al. 1994). The complex transmission can be converted to the conductivity through t ± ( H ) = [4n/(n + 1)]/[Zoda±(H) + n + 1], where t ± is the complex transmission for right- and left-circularly polarized light, H is the magnetic feld, d is the film thickness in cm, Z0 = 377 is the impedance of free space in ohms, and n is the index of refraction of the substrate. 6.1. The ab-plane conductivity in a magnetic field At microwave frequencies, the electrodynamics of HTSC have mostly been discussed in terms of phenomenological models of the response of the vortex system to an external ac field starting with Gittleman and Rosenblum (1966) and more recently by Coffey and Clem (1991). A more microscopic description, including effects of bound states in the vortex core, has been provided by Hsu (1993). For a recent review of the microwave work see Golosovsky et al. (1996). The Lorentz force on a vortex at rest is FL = psq)0[vs × n], where ps is the charge density of the condensate and v~ is the condensate velocity. This force gives the vortex a velocity Vv normal to the current. In the rest frame of the vortex the superfluid now has an additional velocity -Vv giving rise to an additional Lorentz force (the Magnus force) of FL - -psq~0[Vv × n]. The total force on the vortex is the sum of these: FL = psq~0[(Vs - Vv) × n]. To complete the equation of motion for the vortex, a pinning force kr and damping t/ are added, where r is tile position of the vortex. At infrared frequencies the inertia of the vortex comes into play and a vortex mass mv can be included in the equation of motion: psq~0[(Vs - aVv) × n] - t/Vv- k r = mv6v.

(17)

The parameter a has been included to parametrize the strength of the Magnus force. With a = 0, my = 0, we get the simple Gittleman and Rosenblum (GR) model with only two parameters, the viscous drag coefficient t/and the force constant k associated with the vortex pinning. This simple model has been used extensively to fit microwaveloss data in magnetic fields (Golosovsky et al. 1996). There are several mechanisms that give rise to the ~/term. The Bardeen-Stephen mechanism (Bardeen and Stephen 1965) gives ~/= Jvhna)cr, where n and T are the carrier concentration and relaxation rate in the vortex core and COc= eHc2/mc is the cyclotron frequency in a field corresponding to Ho2. A model developed by Coffey and Clem (1991) allows one to calculate t/and k fi'om the measurements of the real and imaginary parts of the microwave surface impedance. Surprisingly, in a large number of experiments on a variety of samples this quantity is found to vary very little from the value of t/ ~ 10-6Ns/m 2, close to the BardeenStephen prediction if a relaxation time of the order of the linearly extrapolated normalstate resistivity is used (Morgan et al. 1994). In contrast, the quasiparticle relaxation rate

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

495

in zero field, as determined from microwave-loss measurements, drops exponentially by several orders o f magnitude just below Tc (Bonn et al. 1992). Similar results were obtained by Parks et al. (1995) using terahertz spectroscopy. The vortex pinning constant k can roughly be estimated by the energy gained as the condensation energy g0H~Z/(8Jv) x ;e~ 2 has to be supplied over the volume o f the vortex if it is to be moved away from the pinning region. Setting this energy equal to the elastic energy k~2/2 o f pinning, one estimates k ~ 0.25/~0Ho2 (Golosovsky et al. 1996). The measured value is 3 × 10 6N/m 2. Solving the Gittelman-Rosenblum equations for the vortex conductivity, one finds a Lorentzian peak centered at zero frequency, with width COo = k/th the so-called depinning frequency. From microwave data where t / a n d k are measured separately, the depinning frequency is found to vary between 25 and 80 GHz. Far-infrared measurements o f thin-film transmission yield a peak at 3.15 c m 1 [with a width of 10 cm -1 (300 GHz), considerably higher than the microwave estimate] (Linh et al. 1996). This is a result o f a much smaller value o f the vortex viscosity ~/, an order of magnitude smaller than the Bardeen-Stephen value o f 1.5x106 (Parks et al. 1995). Parks et al. argue that the small value o f t / c a n be understood in a d-wave picture from a calculation of the d-wave density o f states by Volovik (1993). They suggest that rtas = Bo2Ooon be replaced by r/Dw = (BBo2)i/2~oO,. They also suggest that the vortex stiffness lc o f the phenomenological model may not be the actual pinning parameter, but related to the changes in spectral weight o f the large numbers o f quasiparticles associated with each vortex in the d-wave scenario. To be valid at higher frequencies, the simple RG vortex dynamics model has to be extended. One has to account for the vortex mass as well as excitations of the vortex core. A simple physical picture of these core excitations is a bound state o f a quasiparticle confined to the vortex core of size of the coherence length & This energy is approximately h Q = h2/(2m~ 2) ,-~ A 2 / E F . The microscopic calculation o f the dynamics of vortices has been done by Hsu (1993) and it gives a conductivity for circularly polarized fields,

ine 2

fo
1

(18)

with ~ = (1 - q))g2, a pinning constant aC/) = coo/£2 a, and the vortex lifetime rv. Choi et al. (1994) applied this formula to thin-film transmission experiments in circularly polarized light and found the following parameter values: Tv = 4 0 + 5 cm 1, a = 5 0 ± 2 cm -1. The cyclotron mass was taken to be m = 3. line. They found a strong feature in the T+/T - transmission ratio at 25 cm 1 which they attribute to a hybridized pinning-cyclotron resonance frequency and a vortex-core resonance £2o = 60 cm i. Eldridge et al. (1995) studied the unpolarized reflectance of thicker films than those investigated by Choi et al. and, fitting Hsu's conductivity fixing £20 = 15 cm 1, they find a = 133 cm -1 and 1 / ~ - 197 c m 1.

496

D.N. B A S O V a n d T. T I M U S K

6.2. The c-axis conductivity in a magnetic field Here, measurements have focused on the low-lying plasma resonance, a hybrid mode involving the lattice oscillating antiphase to the electronic system first described by Bonn et al. (1987). The plasma edge has been interpreted in terms of the Josephson plasma resonance resulting from Josephson-pair tunneling between superconducting layers (Basov et al. 1994a). In this model, the relationship between the Josephson plasma frequency COpand the Josephson critical current is given by Bulaevskii (1973):

8:v2 CSjc COp --

(19)

CO (DO '

where d is the interlayer spacing andjc is the Josephson critical current. It was shown by Bulaevskii et al. (1995) that a magnetic field of the order of H0 = Cl)o/ys2 can suppress the critical current and shift the plasma frequency according to COp(B) = COp(O) 1 - g H00 In ~ -

.

For Lal.85 Sr0.15CuO4, this field is of the order of 50 T and the suppression Of Jc is expected to be weak, whereas for Bi2Sr2CaCu2Os+,5 where 7 is of the order of 400, H0 is only of the order of a few tesla. In La~.ssSr0.15CuO4, the c-axis conductivity has been measured in both H II c and H ± c by infrared reflectance spectroscopy on single crystals (Gerrits et al. 1995, van Bentum et al. 1997). In optimally doped Lal.85Sr0.15CuO4, Gerrits et al. found no shift of this plasma resonance with magnetic field (H ± c), while in underdoped Lal ssSr015CUO4, van Bentum et al. (1997) reported a large shift from 22 c m -1 at B = 0 to 12 cm -~ at B = 22 T. The shift was in approximate accord with the theory of Bulaevskii et al. (1995). Similar large shifts have been reported for Bi2 Sr2CaCu2Os + 6 by Tsui et al. (1996) and Matsuda et al. (1995), and for TlzBa2CaCu208 by Dulic et al. (2001). It was found that the frequency of the plasma resonance, which was observed by sweeping the magnetic field at constant microwave frequency, varied as COp2e( B -w, where v ~ 0.7-0.8, depending on the sample. The detailed angular dependence of the plasma resonance showed that as the direction of B came within 5 ° of alignment with the planes, the resonance field displayed a cusp, dropping by 6 T within a few degrees. These effects were predicted by Bulaevskii et al. (1996).

7. Conclusions and open questions Experimental studies of a variety of high-To materials by means of infrared and optical spectroscopies have led to a reasonable consensus on several aspects of the properties of these fascinating materials.

INFRARED PROPERTIESOF HTSC: EXPERIMENTALOVERVIEW

497

Electronic structure. It is clear that the growth of the spectral weight with doping cannot be understood in terms of simple Fermi-liquid models with a large Fermi surface, and that an appropriate starting point for unraveling the exotic behavior of cuprates is the MottHubbard scenario. Experimental results on the evolution of Drude weight with doping are in remarkable agreement with the calculations performed within the MH model. The charge carriers in conducting phases cannot be viewed as free electrons/holes, but should be regarded as correlated "objects" whose properties are strongly influenced by Coulomb effects and possibly by antiferromagnetic fluctuations. An obvious challenge for any model is to explain the qualitative difference between in-plane and interplane properties of the cuprates. This difference goes beyond "strong anisotropy". For example, while the interplane transport is incoherent, at the same time in-plane properties are at least approximately described by a Drude model. Now that there is evidence that the in-plane and interplane charge dynamics are closely interrelated, any realistic description of a system must account for this relationship as well as the dramatic contrasts. Progress in understanding the transport properties. While the principal trends in the growth of the electronic spectral weight with doping were established quite early on, systematic studies of the evolution of charge dynamics in high-Tc materials are still in progress. Nevertheless, several important trends have been revealed by recent experiments. Charge carriers in underdoped cuprates are characterized by a significant enhancement of the mass (up to 5me) and pronounced frequency dependence of the scattering rate. It is likely that interactions giving rise to these properties are of electronic origin and are connected with the proximity of underdoped compounds to parent AF insulators. This latter conclusion is corroborated by the suppression of m* and the frequency-dependent term in 1/r(~o) spectra as one proceeds with doping. In the optimally doped materials, the frequency-dependent scattering rate varies linearly with temperature and energy. In underdoped compounds, there is a stronger suppression of l/r(co) at low frequencies and, at T > T~, this correlates with the opening of a pseudogap as detected by other experimental techniques. The symmetry of the superconducting order parameter, The evidence for strong Coulomb interaction creates formidable difficulties for a complete theory of cuprate superconductivity. One way for two electrons to avoid Coulomb repulsion is to spatially arrange themselves in a wave function of dx2~2 symxnetry (for a review, see Scalapino 1995). This hypothesis ignited an extensive study of possibilities of pairing in d,2~2. Phase-sensitive tunneling probes support this picture (Van Harlingen 1995, Kirtley et al. 1995, Tsuei et al. 1997). Spectroscopic measurements of the penetration depth and of the complex conductivity provide valuable, albeit indirect support for an anisotropic (possibly d-wave) symmetry of the order parameter. The wealth of data on the electromagnetic response is not consistent with s-wave symmetry of the gap. Early experiments claiming s-wave gap are not reproduced with the new generation of high-purity samples which are well characterized by a variety of techniques. As experiments with a number of probes show, the density of states in cuprates at T << Tc is fundamentally different from conventional superconductors. Spectroscopic

498

D.N. BASOVand T. TIMUSK

evidence in favor of a finite density of states in the superconducting state is overwhelming, but it remains to be seen whether this complex experimental situation can be reconciled based solely on d-wave ideas. One of the last remaining "strongholds" for s-wave superconductivity is the response of the electron-doped material Nd2_yCeyCuO4. Further experimental work is required to understand if electron-doped systems are fundamentally different from other cuprates. Some implications for the pairing mechanism. The analysis of the charge dynamics in the normal and the superconducting states which has been carried out for Y123 and Bi2212 compounds suggests that the scattering mechanisms are not due to phonons, but most likely, are related to the spin degree of freedom. Indirectly, this gives support for spin fluctuations being the mediator of superconducting pairing (Pines 1995). This particular pairing interaction gives rise to a d-wave order parameter which is consistent with many experiments (Scalapino 1995). A low-lying pseudogap in the spectrum of spin fluctuations has been shown to enhance the critical temperature of a superconductor (Schachinger et al. 1997). Such a pseudogap is a likely interpretation of the wealth of experimental data obtained for underdoped samples including infrared conductivity of oxygen-deficient Y123, Y124 and Bi2212 (Puchkov et al. 1996a, Timusk and Statt 1999). Open questions. Despite the large existing literature, there are serious gaps in the available data: (i) The low-temperature properties of the single-layered materials are nearly unstudied. By doping, their Tc can be reduced to values comparable to low-Tc BCS materials, and some of the techniques used in low-temperature superconductivity can be applied, such as destruction of superconductivity in a high magnetic field. (ii) Another important issue is related to the properties of overdoped compounds. One of the dogmas in the field is that overdoping "restores Fermi-liquid behavior". There is little experimental data to support this claim. (iii) The large versus small Fermi surface problem. Why does the ab-plane conductivity vary linearly with doping x? The model of a large Fermi surface suggests that it should vary with l - x as the doped holes reduce the size of the Fermi surface. It is hoped that in the near future k-sensitive experiments, such as ARPES and neutron scattering, will provide us with details on the nature of the electronic structure in the low-doping, pseudogap state.

Acknowledgements The authors are indebted to the many individuals who have contributed to the research discussed in this article. In particular we would like to emphasize the impact of our longterm collaborators: D.A. Bonn, W.N. Hardy, A.V. Puchkov, C.C. Homes, J.R Carbotte, D.B. Tanner, G.Blumberg, R. Liang, B. Dabrowski. We also benefitted from the discussions of many problems presented in this review with: A.A. Abrikosov, P.B. Allen, W. Atkinson, A. Balatskii, A.J. Berlinski, J.C. Cooper, O.V Dolgov, R.C. Dynes, V.J. Emery, M. Franz, J. Hirsch, C. Kallin, S.A. Kivelson, RA. Lee, A.G. Loeser, J.W. Loram, M.B. Maple, F. Marsiglio, A.J. Millis, M. Norman, J. Preston, M. Pham, D. Pines, Z.-X. Shen, M. Strongin, T. Startseva, T. R66m, M. Reedyk, S. Tajima, D. van

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW

499

d e r M a r e l , a n d Y. U e m u r a . T h e w o r k at U C S D is s u p p o r t e d b y N S F a n d D O E . T h e w o r k at M c M a s t e r is s u p p o r t e d b y the C a n a d i a n I n s t i t u t e o f A d v a n c e d R e s e a r c h ( C I A R ) a n d Natural Sciences and Engineering Research Council of Canada.

References Abrikosov, A.A., 1995, Phys. Rev. B 52, R7026. Abrikosov, A.A., 1996, Phys. Rev. B 54, 12003. Abrikosov, A.A., and L.A. Fal'kovskii, 1961, Sov. Phys. JETP 13, 179. Alexandrov, A.S., A.M. Bratkovky, N.E Mott and E.K.H. Salje, 1993, Physica C 215, 359. Alexandrov, A.S., VV Kabanov and N.E Mort, 1996, Phys. Rev. Lett. 77, 4799. Allen, R, 1992, Comments Con& Mat. Phys. 5-6, 327. Allen, EB., 1971, Phys. Rev. B 3, 305. Allen, EB., and J.C. Mikkelsen, 1976, Phys. Rev. B 15, 2953. Allen Jr, S.J., D.C. Tsui and E DeRosa, 1975, Phys. Rev. Lett. 35, 1359. Alloul, H., P. Mendels, H. Casalta, J.E Marucco and J. Arabski, 1991, Phys. Rev. Lett. 67, 3140. Altshuller, B.L., L.B. Ioffe and A.J. Millis, 1996, Phys. Rev. B 53, 415. Anderson, P.W, 1958, Plays. Rev. 109, 1492. Anderson, P.V~, 1987, Science 235, 1196. Anderson, P.W., 1995, Science 268, 1154. Arima, T., Y. Tokura and S. Uchida, 1993, Phys. Rev. B 48, 6597. Atkinson, W.A., and J.P. Carbotte, 1995, Phys. Rev. B 52, 10601. Atkinson, WA., and J.P. Carbotte, 1997, Phys. Rev. B 55, 3230. Awasthi, A.M., L. Degiorgi, G. Gruner, Y. Dalichaouch and M.B. Maple, 1993, Phys. Rev. B 48, 10692. Bacci, S.B., E.R. Gagliano, R.M. Martin and J.E Annett, 1991, Phys. Rev. B 44, 7504-7509. Bardeen, J., and M.J. Stephen, 1965, Phys. Rev. 140, Al197. Bardeen, J., L.N. Cooper and LR. Schrieffer, 1957, Phys. Rev. 108, 1175. Barowski, H.S., A. Arnold, R. Eder, M. Prcnninger, A. Vogenaner, E.V Pechen, K.E Renk, W. Prusseit, H. Kinder, R.G. Buckley, D.M. Pooke and K. Kishio, 1996, SPIE 2696, 66. Basov, D.N., T. Timusk, B. Dabrowski and J.D. Jorgensen, 1994a, Phys. Rev. B 50, 351l.

Basov, D.N., A.'~ Puchkov, R.A. Hughes, T. Strach, J. Preston, T. Timusk, R. Liang, D.A. Bonn and W.N. Hardy, 1994b, Phys. Rev. B 49, 12165. Basov, D.N., R. Liang, D.A. Bonn, W.N. Hardy, B. Dabrowski, M.A. Quijada, D.B. Tanner, J.E Rice, D.M. Ginsberg and T. Timusk, 1995a, Phys. Rev. Lett. 74, 595. Basov, D.N., H.A. Mook, B. Dabrowski and T. Timusk, 1995b, Phys. Rev. B 52, R13141. Basov, D.N., R. Liang, B. Dabrowski, D.A. Bonn, WN. Hardy and T. Timusk, 1996, Phys. Rev. Lett. 77, 4090. Basov, D.N., B. Dabrowski and T. Timusk, 1998, Phys. Rev. Lett. 81, 2132. Basov, D.N., ST Woods, A.S. Katz, E.J, Singlcy, R.C. Dynes, M. Xu, D.G. Hinks, C.C. Homes and M. Strongin, 1999, Science 283, 49. Batlogg, B., and VJ. Emery, 1996, Nature 382, 20. Batlogg, B., H.Y. Hwang, H. Takagi, R.J. Cava, H.L. Kao and J. Kwo, 1994, Physica C 235-240, 130. Bauer, M., 1990, Ph.D. Thesis (University of Tfibingen). Bednorz, J.G., and K.A. Mtiller, 1986, Z. Phys. B 64, 189. Bernhard, C., J.L. Tallon, C. Bucci, R. de Renzi, G. Guidi, G.V.M. Williams and Ch. Niedermayer, 1996, Phys. Rev. Lett. 77, 2304. Bernhard, C., R. Henn, A. Wittlin, M. Kl~iser, T. Wolf, G. Mtilter-Vogt, C.T. Lin and M. Cardona, 1998, Phys. Rev. Eett. 80, 1762. Bernhard, C., D. Munzar, A. Witttin, W Konig, A. Golnik, C.T. Lin, M. K1/iser, T. Wolf, G. Mtiller-Vogt and M. Cardona, 1999, Phys. Rev. B 59, R6631. Bernhard, C., D. Munzar, A. Golnik, C.T. Lin, A. Wittlin, J. Humlicek and M. Cardona, 2000, Phys. Rev. B 61, 618. Bi, X.-X., and R Eklund, 1993, Phys. Rev. Lett. 70, 2625. Birmingham, J.T., S.M. Grannan, RL. Richards, R. Hcnn, J. Kircher, M. Cardona, A. Wittlin, V.H.M. Duijn and A.A. Menovsky, 1996, SPIE 2696, 56. Blumberg, G., M. Kang, M.V. Klein, K. Kadowaki and C. Kendziora, 1997, Science 278, 1427.

500

D.N. BASOV and T. T1MUSK

Boebinger, G., Y. Ando, A. Passner, "I: Kimura, M. Okuya, J. Shimoyama, K. Kishio, K. Tamasaku, N. Ichikawa and S. Uchida, 1996, Phys. Rev. Lett. 77, 5417. Bommeli, E, L. Degiorgi, E Wachter, B.K. Cho, RC. Canfield, R. Chau and M.B. Maple, 1997, Phys. Rev. Let*. 78, 547. Bonn, D.A., and W.N. Hardy, 1996, Microwave surface impedance ofhigh-T Csuperconductors, in: Physical Properties of High Temperature Superconductors, Vol. I, ed. D.M. Ginsberg (World Scientific, Singapore). Bonn, D.A., et al., 1987, Phys. Rev B 35, 8843. Bonn, D.A., J.D. Garret and T. Timusk, 1988, Phys. Rev. Lett. 61, 1305. Bonn, D.A., R Dosanjh, R. Liang and W.N. Hardy, 1992, Phys. Rev. Lett. 68, 2390. Bonn, D.A., R. Liang, T.M. Reisman, D.J. Baar, D.C. Morgan, K. Zhang, P. Dosanjh, T.L. Duty, A. MacFarlane, G.D. Morris, J.H. Brewer, WN. Hardy, C. Kallin and A.J. Berlinsky, 1993, Phys. Rev. B 47, 11314. Bonn, D.A., S. Kamal, K. Zhang, R. Liang, D.J. Baar, E. Klein and W.N. Hardy, 1994, Phys. Rev B 50, 4051. Borkowski, L.S., and RJ. Hirschfeld, 1994, Phys. Re,/. B 49, 15404. Born, M., and E. Wolf, 1980, Principles of Optics (Pergamon Press, New York). Bozovic, I., K. Char, S.J.B. Yoo, A. Kapitulnik, M.R. Beasley, T.H. Geballe, Z.Z. Wang, S. Hagen, N.R Ong, D.E. Aspens and M.K. Kelly, 1988, Phys. Rev. B 38, 5077. Bozovic, I., J.H. Kim, J.S. Harris Jr, C.B. Eom, J.M. Phillips and J.T. Cheung, 1994, Phys. Rev. Lett. 73, 1436. Branch, D., and J.R Carbotte, 1998, J. Phys. Chem. Solids 59, 1965. Bucher, B., J. Karpinski, E. Kaldis and R Wachter, 1990, Physica C 167, 324. Bucher, B., J. Karpinski, E. Kaldis and R Wachter, 1992, Phys. Rev. B 45, 3026. Bucher, B., E Steiner, J. Karpinski, E. Kaldis and E Wachter, 1993, Phys. Rev. Lett. 70, 2012. Butaevskii, L.N., 1973, Sov. Phys. JETP 37, 1133. Bulaevskii, L.N., and J.R. Clem, 1991, Phys. Rev. B 44, 10234. Bulaevskii, L.N., M.R Maley and M. Tachiki, 1995, Phys. Rev. Let*. 74, 801. Bulaevskii, L.N., M.R Maley, H. Safar and D. Dominguez, 1996, Phys. Rev. B 53, 6633.

Burlakov, VM., K.V. Kraiskaya, A.G. Mitko and E.J. Firsov, 1989, Phys. Lett. A 142, 514. Calvani, R, M. Capizzi, S. Lupi, R Maselli, A. Paolone and P. Roy, 1996, Phys. Rev. B 53, 2756. Carbotte, J.P., C. Jiang, D.N. Basov and "i2 Timusk, 1995, Phys. Rev. B 51 11798. Carbotte, J.R, E. Schachinger and D.N. Basov, 1999, Nature 401, 354. Chal,a'avarty, S., 1998, Eur. Phys. J. B 5, 337. Chakravarty, S., and RW. Anderson, 1994, Phys. Rev. Let*. 72, 3218. Choi, E.-J., H.-T. Linh, H.D. Drew and T.C. Hsu, 1994, Phys. Rev. B 49, 13271. Choi, E.-J., K.R Stewart, S.K. Kaplan, H.D. Drew, S.N. Mao and T. Venkatesan, 1996, Phys. Rev B 53, 8859. Chubukov, A.V, D. Pines and B.R Stojkovic, 1996, J.Phys.:Condensed Matter 8, 10017. Clarke, DG., S.E Strong and P.W Anderson, 1995, Phys. Rev. Lett. 74, 4499. Coffey, M.W., 1995, Phys. Lett. A 200, 195. Coffey, M.W., and J.R. Clem, 1991, Phys. Rev. Lett. 67, 386. Collins, R.T., Z. Schlesinger, E Holtzberg, E Chaudhari and C. Feild, 1989, Phys. Rev. Let*. 39, 6571. Combescot, R., and X. Leyronas, t995, Phys. Rev. Lett. 75, 3732. Cooper, J.R., 1996, Phys. Rev. B 54, R3753. Cooper, J.R., L. Form and B. Keszei, 1990, Nature 343, 444. Cooper, S.L., G.A. Thomas, J. Orensten, D.H. Rapkine, A.J. Millis, S.-W. Cheong, A.S. Cooper and Z. Fisk, 1990, Phys. Rev. B 41, 11605. Cooper, S.L., A.L. Kotz, M.A. Karlow, M.V Klein, W.C. Lee, J. Giapintzakis and D.M. Ginsberg, 1992, Phys. Rev. B 45, 2549. Cooper, S.L., D. Reznik, A.L. Kotz, M.A. Karlow, R. Liu, M.V. Klein, W.C. Lee, J. Giapintzakis, D.M. Ginsberg, B.W. Veal and A.R Paulikas, t993a, Phys. Rev. B 47, 8233. Cooper, S.L., R Nyhus, D. Reznik, M.V Klein, W.C. Lee, D.M. Ginsberg, B.W. Veal, A.R Paulikas and B. Dabrowski, 1993b, Phys. Rev. Lett. 70, 1533. Dagotto, E., 1994, Rev. Mod. Phys. 66, 763. Dahne, U., Y. Goncharov, N. Klein, N. Tellmann, G. Kozlov and K. Urban, 1995, J. Superconductivity 8, 129. de Vaulchier, L.A., J.R Vieren, Y. Guldner, N. Bontemps, R. Combescot, Y. Lemaitre and J.C. Mage, 1996, Europhys. Lett. 33, 153.

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW Degiorgi, L., E.J. Nicol, O. Klein, G. Grunner, E Wachter, S.-M. Huang, J. Wiley and R.B. Kaner, 1994, Phys. Rev. B 49, 7012. Dewing, H.L., and E.K.H. Salje, 1992, Supercond. Sci. Technol. 5, 50. Ding, H., T. Yokoya, J.C. Campuzano, T. Takahashi, M. Randeria, M.R. Norman, T. Mochiku, K. Kadowaki and J. Giapintzakis, 1996, Nature 382, 51. Ding, H., M.R. Norman, T. Yokoya, T. Takeuchi, M. Randeria, J.C. Campuzano, T. Takahashi, T. Mochiku and K. Kadowaki, 1997, Phys. Rev. Lett. 78, 2628. Dolgov, O., HJ. Kaufmann, E.K.H. Salje and Y. Yagil, 1997, Physica C 279, pl13. Dutic, D., S.J. Hak, D. van der Marel, W.N. Hardy, A.E. Koshelev, R. Liang, D.A. Bonn and B.A. Willemsen, 2001, Phys. Rev. Lett. 86, 4660. Edwards, H.L., D.J. Derro, A.L. Barr, J.T. Markert and A.L. de Lozanne, 1995, Phys. Rev. Lett. 75, 1388. E1 Azrak, A., R. Nahoum, N. Bontemps, M. GuillouxViry, C. Thivet, A. Perrin, S. Labdi, Z.Z. Li and H. Raffy, 1994, Phys. Rev. B 49, 9846. Eldridge, J.E., M. Dressel, D.J. Matz, B. Gross, Q.Y. Ma and W.N. Hardy, 1995, Phys. Rev B 52, 4462. Emery, VJ., and S.A. Klvelson, 1995a, Phys. Rev. Lett. 74, 3253. Emery, V.J., and S.A. Kivelson, 1995b, Na_ture 374, 434. Emery, VJ., S.A. Kivelson and O. Zachar, 1997, Phys. Rev. B 56, 6120. Eskes, H., M.B.J. Meinders and G.A. Sawatzky, 1991, Phys. Rev. Lett. 67, 1035. Falck, J.E, A. Levy, M.A. Kastner and R.J. Birgeneau, 1992, Plays. Rev. Lett. 69, 1109. Fano, U., 1961, Phys. Rev 121, 1866. Fawcett, E., 1988, Rev. Mod. Phys. 60, 209 283. Fehrenbacher, R., and M.R. Norman, 1994, Plays. Rev. B 50, 3495. Fong, H.E, B. Keimer, P.W. Anderson, D. Reznik, E Dogan and I.A. Aksay, 1995, Phys. Re,/. Lett. 75, 316. Fong, H.E, B. Keimer, D.L. Milius and I.A. Aksay, 1997, Phys. Rev. Lett. 78, 713. Forro, L., G.L. Cart, G.P. Williams, D. Mandrus and L. Mihaly, 1990, Phys. Rev. Lett. 65, 1941. Franz, M., C. Kallin, A.J. Berlinsky and M.I. Salkola, 1997, Phys. Rev. B 56 7882. Friedmann, T.A., M.W Rabin, J. Giapintzakis, J.P. Rice and D.M. Ginsberg, 1990, Phys. Rev. B 42, 6217. Fnjimori, A., E. Takayama-Muromachi, Y. Uchida and B. Okai, 1987, Phys. Rev. B 35, 8814.

501

Fukuzumi, Y., K. Mizuhashi, K. Takenaka and S. Uchida, 1996, Phys. Rev. Lett. 76, 684. Gao, E, D.B. Romero, D.B. Tanner, J. Talvacchio and M.G. Forrester, 1993a, Phys. Rev. B 47, 1036. Gao, E, J,W. Kruse, C.E. Platt, M. Feng and M.M Klein, 1993b, Appl. Phys. Lett. 63, 2274. Gauzzi, A., and J. Bok, 1996, J. Superconductivity 9, 523. Georges, A., G. Kotliar, W. Krauth and M.J. Rozenberg, 1996, Rev. Mod. Phys. 68, 13. Gerrits, A.M., et al., 1994, Physica C 235-240, 1117. Gerrits, A.M., M.E.J. Boonman, A. Wittlin, EJ.M. van Bentum, V.H.M. Duijn and A.A. Menovsky, 1995, Phys. Rev. B 51, 12049. Gervais, E, E Echegut, LM. Bassat and E Odier, 1988, Phys. Rev. B 37, 9364. Gittleman, J.I., and B. Rosenblum, 1966, Phys. Rev. Lett. 16 734. Glover, R.E., and M. Tinkham, 1956, Phys. Rev. 104, 844. Gold, A., S.J. Allen, B,A. Wilson and D.C. Tsui, 1982, Phys. Rev. B 25, 3519. Golosovsky, M., M. Tsindlekht and D. Davidov, 1996, Supercond. Sci. Technol. 9, 1. G6tze, W., 1981, Philos. Mag. B 43, 219. G6tze, W., andP. W61fle, 1972, Phys. Rev. B 6, 1226. Graf, M.J., D. Rainer and J.A. Sauls, 1995a, Phys. Rev. B 47, 12089. Graf, M.J., M. Palumbo and D. Rainer, 1995b, Phys. Re,/. B 52, 10588. Grishkovsky, D., S. Keiding, M. van Exter and Ch. Fattinger, 1990, J. Opt. Soc. Am. 7, 2006. Grtininger, M., J. Mtinzel, A. Gaymann, A. Zibold and H.P. Geserich, 1996, Europhys. Lett. 35, 55. Grtiinnger, M., D. van der Marel, A.A. Tsvetkov and A. Erb, 2000, Phys. Rev. Lett. 84, 1575. Gurvitch, M., and A.T. Fiory, 1987, Phys. Rev. Lett. 59, 1337. Hackl, R., G. Krug, R. Nemetschek, M. Opel and B. Stadlober, 1996, in: Spectroscopic Studies of Superconductors, Vol. V, eds I. Bozovic and D. van der Marel, SPIE Proc. 2696, 194. Hardy, W.N., D.A. Bonn, D.C. Morgan, R. Liang and K. Zhang, 1993, Phys. Rev. Lett. 70, 3999. Hardy, W.N., S. Kamal, R. Liang, D.A. Bonn, C.C. Homes, D.N. Basov and T. Timusk, 1996, in: Proc. 10th Anniversary of High-T c Workshop (World Scientific, Singapore) p. 223. Harris, J.M., EJ. White, Z.-X. Shen, H. Ikeda, R. Yoshizald, H. Eisaki, S. Uchida, W.D. Si,

502

D.N. BASOV and T. TIMUSK

J.W. Xiong, Z.-X. Zhao and D.S. Dessau, 1997, Phys. Rev. Lett. 79, 143. Hauff, R., S. Tajima, W.J. Jang and A.I. Rykov, 1996, Plays. Rev. Lett. 77 4620. Henn, R., J. Kircher, M. Cardona, A. Wittlin, VH.M. Duijn and A.A. Menovsky, 1996, Phys. Rev. B 53, 9353. Hirsch, J.E., 1992, Physica C 199, 305. Hirschfeld, EJ., E W61fle, J.A. Sauls, D. Einzel and W.O. Putikka, 1989, Phys. Rev. B 40, 6695. Hirschfeld, EL, W.O. Putikka and D.J. Scalapino, 1994, Phys. Rev. B 50, 10250. Holmes, W, and EL. Riehards, 1997, unpublished. Homes, C.C., M. Reedyk, D.A. Crandles and T. Timusk, 1993a, Appl. Opt. 32, 2976. Homes, C.C., 32 Timusk, R. Liang, D.A. Bonn and W.N. Hardy, 1993b, Phys. Rev. Lett. 71, 1645. Homes, C.C., T. Timusk, D.A. Bonn, R. Liang and W.N. Hardy, 1995a, Physica C 254, 265. Homes, C.C., "i2 Timusk, D.A. Bonn, R. Liang and V~N. Hardy, 1995b, Can. J. Phys. 73, 663. Homes, C.C., B.R Clayman, J.L. Peng and R.L. Greene, 1997, Phys. Rev. B 56, 5525. Homes, C.C., D.A. Bonn, R. Liang, W.N. Hardy, D.N. Basov, T. Timusk and B.P. Clayman, 1999, Phys. Rev B 60, 9782. Hsu, T.C., 1993, Physica C 231, 305; Plays. Rev. B 46, 3680. Hybertsen, M.S., E.B. Stechel, M. Schlnter and D.R. Jennison, 1990, Phys. Rev. B 41, 11068. loffe, L.B., A.I. Larkin, A.A. Varlamov and L. Yu, 1993, Phys. Rev. B 47, 8936. Ioffe, L.B., A.I. Larkin, A.J. Millis and B.L. Altshuller, 1994, JETP Lett. 59, 66. lto, %, K. Takenaka and S. Uchida, 1993, Phys. Rev. Lett. 70, 3995. Jacobs, T., S. Sridar, Q. Li, G.D. Gu and N. Koshizuka, 1995, Phys. Rev. Lett. 75, 4516. Jang, H.E, G. Cripps and T. Timusk, 1990, Phys. Rev. B 41, 5152. Jarell, M., J.K. Freeficks and Th. Pruschke, 1995, Phys. Rev. B 51, 11704. Jiang, C., E. Schachinger, J.R Carbotte, D.N. Basov and T. Timusk, 1996, Phys. Rev. B 54, 1264. Kajueter, H., G. Kotliar and G. Moeller, 1996, Phys. Rev. B 53 16214. Kamai, S., R. Liang, A. Hosseini, D.A. Bonn and W.N. Hardy, 1998, Phys. Rev. B 56, R8933. Kamarfis, K., S.L. Herr, C.D. Porter, N. Tache, D.B. Tanner, S. Etemad, 32 Venkatesan, E. Chase, A.

Inam, X.D. Wu, M.S. Hegde and B. Dutta, 1990, Phys. Rev. Lett. 64, 84. Kampf, A.P, and J.R. Schrieffer, 1990, Phys. Rev. B 42, 7967. Kastner, M.A., and R.J. Birgeneau, 1996, SHE 2696, 45. Katsufuji, T., Y. Okimoto and Y. Tokm'a, 1995, Phys. Rev. Lett. 75, 3497. Katsufuji, 32, T. Tanabe, 32 Ishikawa, Y. Fukuda, T. Arima and Y. Tokura, 1996, Phys. Rev. B 54, R14230. Keimer, B., W.Y. Shih, R.W. Erwin, J.W. Lynn, E Dogan and I.A. Aksay, 1994, Phys. Rev. Lett. 73, 3459. Kim, J.-T., T.R. Lemberger, S.R. Foltyn and X. Wu, 1994, Phys. Rev. B 49, 15970. Kircher, J., M. Cardona, A. Zibold, H.P Geserich, E. Kaldis, J. Karpinski and S. Rusieeki, 1993, Phys. Rev. B 48, 3993. Kircher, J., D. Miller, P.L. Richards, E. Kaldis and J. Karpinski, 1994, Physica C 235-240, 1119. Kircher, J., R. Henn, M. Cardona, EL. Richards and G.R. Williams, 1997, J. Opt. Soc. Am. B 14, 705. Kirtley, LR., C.C. Tsuei, J.Z. Sun, C.C. Chi, L.S. YuJahnes, A. Gupta, M. Rupp and M.B. Ketchen, 1995, Nature 373, 225. Kirtley, J.R., K.A. Moler, G. Villard and A. Maignan, 1998, Phys. Rev. Lett. 81, 2140. Klemm, R.A., and S.H. Liu, 1995, Phys. Rev. Lett. 74, 2343. Koch, B., H.P Geserich and Th. Wolf, 1989, Solid State Comm. 71,495. Koller, D., et al., 1996, Phys. Rev. Lett. 77, 4082. Krasnov, V~M., A. Yurgens, D. Win!tier, P Delsing and T. Claeson, 2000, Phys. Rev. Lett. 84, 5860. Kresin, MZ., and S.A. Wolf, 1992, Phys. Rev. B 46, 6458. Kumar, N., and A.M. Jayannavar, 1992, Phys. Rev. B 50, 438. Kumar, N., PA. Lee and B. Shapiro, 1990, Physica A 168, 447. Kwak, J.E, 1982, Phys. Rev. B 26, 4789. Landau, L.D., and E.M. Lifshitz, 1992, Statistical Physics (Pergamon Press, Oxford). Laughlin, R.B., 1996, Phys. Rev. Lett. 79, 1726. Lawrence, W.E., and S. Doniach, 1971, in: Proc. 12th Int. Conf. on Low Temperature Physics, ed. E. Kanda (Academic Press, New York) p. 361. Lee, PA., 1993, Phys. Rev. Lett. 71, 1887. Lee, RA., and N. Nagaosa, 1992, Phys. Rev. B 46, 5621.

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW Linh, H.-T., S. Wu, H.D. Drew, S.K. Kaplan, Q. Li and D.B. Fenner, 1996, Phys. Rev. Lett. 76, 3810. Lippert, M., J.R Strobel, G. Saemann-lshenko, S. Orbach, S. Hensen, G. Muller, H. Piel, J. Schi~tzmann, K.E Renk, B. Roas and V~ Gieres, 1991, Physica C 185-189, 1041. Litttewood, RB., and C.M. Varma, 1992, Phys. Rev. B 46, 405. Litvinchuk, A.P., C. Thomsen and M. Cardona, 1994, in: Physical Properties of High Temperature Superconductors, Vol. IV,,ed. D.M. Ginsberg (World Scientific, Singapore). Liu, Y., and A.M. Goldman, 1994, Mod. Phys. Lett. 8, 277. Lobo, R.ES.M., EJ. Gotor, P Odier and E Gervais, 1996, Phys. Rev. B 53, 410. Loeser, A.G., Z.-X. Shen, D.S. Dessau, D.S. Marshall, C.-H. Park, E Fournier and A. Kapitulnik, 1996, Science 273, 325. Loram, J.W., K.A. Mirza, J.R. Cooper and W.Y. Liang, 1994, J. Superconductivity 7, 261. Lorenzana, J., and G.A. Sawatzky, 1995, Phys. Rev. Lett. 74, 1867. Lyons, K.B., RA. Fleury, J.R Remeika, A.S. Cooper and T.J. Negran, 1988, Phys. Rev. B 37, 2353. Ma, Z., 1995, Ph.D. Thesis (Stanford University). Mahan, G.D., 1980, Many-Particle Physics (Plenum Press, New York). Maksimov, E.G., S.N. Rashkeev, S.Yu. Savrasov and Yu.A. Uspenski, 1989, Phys. Rev. Lett. 63, 1880. Man&us, D., M.C. Martin, C. Kendziora, D. Koller, L. Forro and L. Mihaly, 1993, Phys. Rev. Lett. 70, 2629. Mao, J., D.H. Wu, J.L. Peng, R.L. Greene and S.M. Anlage, 1995, Phys. Rev. B 51, 3316. Marshall, D.S., D.S. Dessau, A.G. Loeser, C.H. Park, Z.-X. Shen, A.Y. Matsuura, J.N. Eckstein, I. Bozovic, P. Fournier, A. Kapitulnik, W.E. Spicer and Z.-X. Shen, 1996, Phys. Rev. Lett. 76, 4841. Marsiglio, E, and J.R Carbottc, 1997, Can. J. Phys. 75, 509. Matsuda, Y., M.B. Gaifullin, K. Kumagai, K. Kadowaki and T. Mochiku, 1995, Phys. Rev. Lett. 75, 4512. Mattis, D.C., and J. Bardeen, 1958, Phys. Rev. 111, 412. McGuire, JJ., M. Windt, T. Startseva, T. Timusk, D. Colson and V. Viallet-Guillen, 2000, Phys. Rev B 62, 8711. Meinders, M.B.J., H. Eskes and G.A. Sawatzky, 1993, Phys. Rev. B 48, 3916.

503

Mihailovic, D., C.M. Foster, K. Voss and A.J. Heeger, 1990, Phys. Rev. B 42, 7989. Miller, D., EL. Richards, S. Etemad, A. Inam, T. Venkatesan, B. Dutta, X.D. Wu, C.B. Eom, T.H. Geballe, N. Newman and B.E Cole, 1993, Phys. Rev. B 47, 8076. Miller, D., RL. Richards and E Merchant, 1995, Phys. Rev. B 51, 8385. Millis, A.J., and H. Monien, 1996, Phys. Rev. B 54, 16172. Moffat, S., 1997, Ph.D. Thesis (McMaster University). Moler, K.A., J.R. Kirtley, D.G. Hinks, T.W. Li and Ming Xu, 1998, Science 279, 1193. Mook, H.A., M. Yethiraj, G. Aeppli, T.E. Mason and T. Armstrong, 1993, Phys. Rev. Lett. 70, 3490. Morgan, D.C., et al., 1994, Physica C 235-240, 2015. Mori, H., 1965, Prog. Theor. Phys. 34, 399. Motohashi, T., J. Shimoyama, K. Kitazawa, K. Kishio, K.M. Kojima, S. Uchida and S. Tajima, 2000, Phys. Rev B 61, R9269. Munzar, D., C. Bernhard, A. Golnik, J. Hurnlicek and M. Cardona, 1999, Solid State Commun. 112,365. Mtinzel, J., A. Zibold, H.P. Geserich and Th. Wolf,. 1996, Europhys. Lett. 33, 147. Nachumi, B., A. Keren, K. Kojima, M. Larkin, G.M. Luke, J. Merrin, O. Tchernyshov, Y.J. Uemura, N. Ichikawa, M. Goto and S. Uchida, 1996, Phys. Rev. Lett. 77, 5421. Naeini, J.G., X.K. Chen, K.C. Hewitt, J.C. Irwin, T.R Devereaux, M. Okuya, T. Kimura and K. Kishio, 1998, Phys. Rev. B 57, Rl1077. Nakamura, Y., and S. Uchida, 1993, Phys. Rev. B 47, 8369. Nemetschek, R., M. Opel, C. Hoffmann, RE Mfiller, R. Hackl, H. Berger, L. Forro, A. Erb and E. Walker, 1997, Phys. Rev. Lett. 78, 4837. Ng, H.K., M. Capizzi, G.A. Thomas, R.N. Bhatt and A.C. Gossard, 1986, Phys. Rev. Lett. 33, 7329. Norman, M.R., H. Ding, T. Yokoya, T. Takeuchi, M. Randeria, J.C. Campuzano, T. Takahashi, T. Mochikn and K. Kadowaki, 1997, Phys. Rev. Lett. 79, 3506. Nyhus, P., M.A. Karlow, S.L. Cooper, B. Veal and A.R Paulikas, 1994, Phys. Rev. B 50, 13898. Orenstein, J., G.A. Thomas, A.J. Millis, S.L. Cooper, D.H. Rapkine, T. Timusk, L.F. Schneemeyer and J.V. Waszczak, 1990, Phys. Rev. B 42, 6342. Osafune, T., N. Motoyama, H. Eisaki and S. Uchida, 1997, Phys. Rev. Lett. 78, 1980.

504

D.N. BASOV and T. TIMUSK

Palumbo, M., and M.J. Graf, 1996, Phys. Rev. B 53, 2261. Panagopoulos, C., J.R. Cooper, N. Athanassopoulou and J. Chrosch, 1996, Phys. Rev. B 54 12721. Parks, B., S. Spielman, J. Orenstein, D.T. Nemeth, E Ludwig, J. Clarke, R Merchant and D.J. Lew, 1995, Phys. Rev. Left. 74, 3265. Perkins, J.D., J.M. Graybeal, M.A. Kastner, R.J. Birgeneau, J.E Falck and M. Greven, 1993, Phys. Rev. Lett. 71, 1621. Perkins, J.D., D.S. Klenberg, M.A. Kastner, R.J. Birgeneau, Y. Endoh, K. Yamada and S. Hosoya, 1995, Phys. Rev. B 52, R9863. Petrov, M.R, A.I. Grachev, M.V Krasnikova, A.A. Nechitailov, V.V. Prokofev, VV Poborchy, S.I. Shaigin and N.E Kartenko, 1988, Solid State Comm. 67, 1197. Pham, 32, H.D. Drew, S.H. Moseley and J.Z. Lin, 1990, Phys. Rev. B 41, 11681. Pham, 32, M.W. Lee, H.D. Drew, U. Welp and Y. Fang, 1991, Phys. Rev. B 44, 5377. Pickett, W.E., 1989, Rev. Mod. Phys. 61,433. Pickett, W.E., R.E. Cohen and H. Krakaner, 1990, Phys. Rev. B 42, 8764. Pickett, W.E., H. Krakauer, R.E. Cohen and D.J. Singh, 1992, Science 255, 46. Pines, D., 1995, in: High Temperature Superconductivity and the C60 Family, ed. H.C. Ren (Gordon and Breach, New York) p. 1. Pines, D., and R Nozibres, 1966, The Theory of Quantum Liquids (Benjamin, New York). Preosti, G., H. Kim and P Muzikal, 1994a, Phys. Rev. B 50, 1259. Preosti, G., H. Kim and R Muzikar, 1994b, Phys. Rev. B 50, 13638. Prohammet, M., and J.E Carbotte, 1991, Phys. Rev. B 43, 5370. Puchkov, A.V., T. Timusk, S. Doyle and A.M. Herman, 1995, Phys. Rev. B 51, 3312. Puchkov, A.V., D.N. Basov and T. Timusk, 1996a, J.Phys.:Condensed Mat. 8, 10049. Puchkov, A.V, R Fournier, 32 Timusk and N.N. Kolesnikov, 1996b, Phys. Rev. Lett. 77, 1853. Puchkov, A.V, R Fournier, D.N. Basov, 32 Timusk, A. Kapitulnik and N.N. Kolesnikov, 1996e, Plays. Rev. Lett. 77, 3212. Puchkov, A.V, T. Timusk, M.A. Karlow, S.L. Cooper, ED. Han and D.A. Payne, 1996d, Phys. Rev. B 54, 6686. Quijada, M.A., D.B. Tanner, R.J. Kelly and M. Oriellion, 1994, Z. Phys. B 94, 255.

Quijada, M.A., D.B. Tanner, EC. Chou, D.C. Johnston and S.-W. Cheong, 1995, Phys. Rev. B 52, 15485. Quinlan, S.M., EJ. Hirschfeld and D.J. Scalapino, 1996, Phys. Rev. B 53, 8575. Radtke, R.L, K. Levin, H.-B. Shuttler and M.R. Norman, 1993, Plays. Rev. B 46, 11975. Reedyk, M., 1992, Ph.D. Thesis (McMaster University). Reedyk, M., and T. Tinnrsk, 1992, Phys. Rev. Left. 69, 2705. Reedyk, M., D.A. Bonn, J.D. Garrett, J.E. Greedan, C.V Stager, T. Timusk, K. Kamarfis and D.B. Tanner, 1988, Phys. Rev. B 38, 11981. Reedyk, M., T. Timusk, J.S. Xue and J.E. Greedan, 1994, Phys. Rev. B 49, 15984. Renk, K.E, B. Gorshunov, J. Sehfitzmann, A. Prtikl, B. Brunner, J. Betz, S. Orbach, N. Klein, G. Mfiller and H. Piel, 1991, Europhys. Lett. 15, 661. Remler, Ch., B. Revaz, J.-Y. Genoud, K. Kadowaki and O. Fischer, 1998, Phys. Rev. Lett. 80, 149. Rice, J.R, J. Giapintzakis, D.M. Ginsberg and J.M. Mochel, 1991, Phys. Rev. B 44, 10158. Roddick, E., and D. Stroud, 1995, Phys. Rev. Lett. 74, 1430. Rojo, A.G., and K. Levin, 1993, Phys. Rev. B 48, 16861. Romero, D.B., C.D. Porter, D.B. Tanner, L. Forro, D. Mandrus, L. Mihaly, G.L. Carr and G.R Williams, 1992, Phys. Rev. Left. 68, 1590. Rossat-Mignod, J., L.E Regnault, C. Vettier, Ph. Bourges, P Burlet, J. Bossy, J.Y. Henry and G. Lapertot, 1991, Physica C 185 189, 86. Rotter, L.D., Z. Schlesinger, R.32 Collins, E Holtzberg, C. Feild, U. Welp, G.W. Crabtree, J.Z. Liu, Y. Fang, K.G. Vandervoort and S. Fleshier, 1991, Phys. Rev. Lett. 67, 2741. Rozenberg, M.J., G. Kotliar, H. Kajueter, G.A. Thomas, D.H. Rapkine, J.M. Honig and E Metealf; 1995, Phys. Rev. Lett. 75, 105. Ruvalds, J., 1996, SPIE 2696, 109. Salkola, M.I., VJ. Emery and S.A. Kivelson, 1996, Phys. Rev. Lett. 77, 155. Scalapino, D.J., 1995, Phys. Rep. 250, 329. Schachinger, E., J.R Carbotte and E Marsiglio, 1997, Phys. Rev. B 56, 2738. Schlesinger, Z., R.T. Collins, E Holtzberg, C. Feild, S.H. Blanton, U. Welp, G.W. Crabtree, Y. Fang and J.Z. Liu, 1990, Phys. Rev. Lett. 65, 801. Schfitzmann, J., B. Gorshunov, K.E Renk, J. Miinzel, A. Zibold, H.E Geserich, A. Erb and G. MfillerVogt, 1992, Phys. Rev. B 46, 512.

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW Schtitzmann, J., S. Tajima, S. Miyamoto and S. Tanaka, 1994, Phys. Rev. Lett. 73, 174. Schtitzmann, J., H.S. Somal, A.A. Tsvetkov, D. van der Marel, G.E.J. Koops, N. Kolesnikov, Z.E Ren, J.H. Wang, E. Bruk and A.A. Menovsky, 1997, Phys. Rev. B 55, 11118. Shastry, B.S., and B.I. Shraiman, 1990, Phys. Rev. Lett. 65 1068. Shen, Z.-X., and J.R. Schrieffer, 1997, Phys. Rev. Lett. 78, 1771. Shibata, H., K. Semba, A. Matsuda and T. Yamada, 1995, Phys. Rev. B 51, 9294. Shibauchi, T., H. Kitano, K. Uchinokura, A. Maeda, T. Kimura and K. Kishio, 1994, Phys. Rev. Lett. 72, 2263. Shulga, S.V, O.V Dolgov and E.G. Maksimov, 1991, Physica C 178, 266. Sigrist, M., and K. Ueda, 1991, Rev. Mod. Phys. 63, 239. Slakey, F., M.V Klein, J.E Rice and D.M. Ginsberg, 1990, Phys. Rev. B 42 2643. Smith, D.Y., 1985, Dispersion theory, sum rules, and their application to the analysis of optical data, in: Handbook of Optical Constants of Solids, ed. E.D. Palik (Academic Press, Orlando, FL). Somal, H.S., B.J. Feenstra, J. Schtitzmann, J.H. Kim, Z.H. BarbeI, VH.M. Duijn, N.T. Hien, A.A. Menovsky, M. Palumbo and D. van der Marel, 1996, Phys. Rev. Lett. 76, 1525. Spielman, S., B. Parks, J. Orenstein, D.T. Nemeth, E Ludwig, J. Clarke, E Merchant and D.J. Lew, 1994, Phys. Rev. Lett. 73, 1537. Srikanth, H., B.A. Willemsen, T. Jacobs, S. Sridar, A. Erb, E. Walker and R. Fltikiger, 1997, Phys. Rev. B 55, R14733. Stadlober, B., G. Krug, R. Nemetschek, R. Hackl, J.L. Cobb and J.T. Markert, 1995, Phys. Rev. Lett. 74 4911. Startseva, T., T. Timusk, A.V. Puchkov, D.N. Basov, H. Mook, M. Okuya, T. Kimura and K. Kishio, 1999a, Phys. Rev. B 59, 7184. Startseva, T., T. Tirnusk, M. Okuya, T. Kimura and K. Kishio, 1999b, Physica C, 321, 135. Stephan, W., and E Horsch, 1990, Phys. Rev. B 42, 8736. Stojkovic, B.R, and D. Pines, 1996, Phys. Rev. Lett. 76, 811. Sulewski, RE., A.J. Sievers, M.B. Maple, M.S. Torikachvili, J.L. Smith and Z. Fisk, 1988, Phys. Rev. B 38, 5338.

505

Sumner, M.J., J.-T. Kim and T.R. Lemberger, 1993, Phys. Rev. B 47, 12248. Sun, A.G., L.M. Patflius, D.A. Gajewski, M.B. Maple and R.C. Dynes, 1994, Phys. Rev. B 50, 3266. Sun, A.G,, S.H. Han, A.S. Katz, D.A. Gajewski, M.B. Maple and R.C. Dynes, 1995, Phys. Rev. B 52, R15731. Tajima, S., G.D. Gu, S. Miyamoto, A. Odagawa and N. Koshizuka, 1993, Phys. Rev. B 48, 16164. Tajima, S., J. Schtitzmarm, S. Miyamoto, I. Terasaki, Y. Sato and R. Hauff, 1997, Phys. Rev. B 55, 6051. Tajima, S., N.L. Wang, R. Hauff, A.I. Rykov and T. Mimnra, 1998, J. Phys. Chem. Solids 59, 2015. Tajima, S., T. Noda, H. Eisaki and S. Uchida, 2001, Phys. Rev. Lett. 86, 500. Takenaka, K., and S. Uchida, 1994, Phys. Rev. B 50, 6534. Tallon, J.L., C. Bernhard, U. Binninger, A. Hofer, G.V.M. Williams, E.J. Ansatdo, J.I. Budnick and Ch. Niedermayer, 1995, Phys. Rev. Lett. 74, 1008. Tamasaku, K., Y. Nakamura and S. Uchida, 1992, Phys. Rev. Lett. 69, 1455. Tanner, D.B., and T. Timusk, 1992, Infrared Properties of High T CSuperconductors, in: Physical Properties of High Temperature Superconductors, Vol. III, ed. D.M. Ginsberg (World Scientific, Singapore). Tanner, D.B., E Gao, M.A. Quijada, D.B. Romero, J.P Rice, D.M. Ginsberg, J. Talivacchio, M.G. Forrester, L. Forro, D. Mandrus, L. Mihaly, G.L. Can and G.R Williams, 1992, J. Phys. Chem. Solids 53, 1611. Tanner, D.B., Y. Yoon, A. Zibold, H.L. Liu, M.A. Quijada, S.W Moore, J.M. Graubeal, B.H. O, J.T. Market, R.J. Kelley, M. Onellion and J.-H. Cho, 1996, SPIE 2696, 13. Tao, H.J., Fanm Lu and E.L. Wolf, 1997, Physica C 282 287, 1507. Thomas, G.A., 1991, in: Proc. 39th Scottish Universities Summer School in Physics of HighTemperature Superconductivity, eds D.P. Tunstall and W Barford (Adam Hilger, Bristol) p. t69. Thomas, G.A., J. Orenstein, D.H, Rapkine, M. Capizzi, A.J. Millis, R.N. Bhatt, L.E Schneemeyer and J.V Waszczak, 1988, Phys. Rev. Lett. 61, 1313. Thomas, G.A., D.H. Rapkine, S.L. Cooper, S.W. Cheong and A.S. Cooper, 1991, Phys. Rev. Lett. 67, 2906. Thomas, G.A., D.H. Rapkine, S.L. Cooper, S.W Cheong, A.S. Cooper, L.E Schneemeyer and J.V. Waszczak, 1992, Phys. Rev. B 45, 2474.

506

D.N. BASOV and T. TIMUSK

Thomas, G.A., D.H. Rapkine, S.A. Carter, A.J. Millis, T.E Rosenbaum, P. Metcalf and J.M. Honig, 1994, Phys. Rev. Lett. 73, 1529. Thomsen, C., and M. Cardona, 1989, in: Physical Properties of High Temperature Superconductors, Vol. I, ed. D.M. Ginsberg (World Scientific, Singapore). Timusk, T., and B.W Statt, 1999, Rep. Progr. Phys. 62, 61. Timusk, T., and D.B. Tanner, 1989, infrared properties of high T,. superconductors, in: Physical Properties of High Temperatm'e Superconductors, Vol. I, ed. D.M. Ginsberg (World Scientific, Singapore) p. 339. Timusk, T., C.C. Homes and W. Reichardt, 1995a, in: Proc. Int. Workshop on Anharmonic Properties of High-T,: Cuprates, Bled, Slovenia, September 1994, eds D. Mihailovic, G. Ruani, E. Kaldis and K.A. Miiller (World Scientific, Singapore) p. 171. Timusk, T., D.N. Basov, C.C. Homes, A.V Puchkov and M. Reedyk, 1995b, J. Superconductivity 8, 437. Tinkham, M., and R.A. Ferrell, 1959, Phys. Rev. Lett. 2, 331. Tokura, Y., S. Koshihara, T. Arima, H. Takagi, S. Ishibashi, T. Ido and S. Uchida, 1990, Phys. Rev. B 41, 11657. Tolpygo, S.K., J.-Y. Lin, M. Gurvitch, S.Y. Hou and J. Phillips, 1996, Phys. Rev. B 53, 12454. Tranquada, J.M., RM. Gehring, G. Shh-ane, S. Shamoto and M. Sato, 1992, Phys. Rev. B 46, 5561. Tsuei, C.C., J.R. Kirtley, Z.F. Ren, J.H. Wang, H. Raffy and Z.Z. Li, 1997, Nature 238, 481. Tsui, O.K.C., N.R Ong and J.B. Peterson, 1996, Phys. Rev. Lett. 76, 819. Tsvetkov, A.A., et al., 1998, Nature 395, 360. Uchida, S., 1997, Hyperfine Interactions 105, 47. Uchida, S., T. Ido, H. Takagi, T. Arima, Y. Tokura and S. Tajima, 1991, Phys. Rev. B 43, 7942. Uchida, S., K. Tamasaku and S. Tajima, 1996, Phys. Rev. B 53, 1. Uemura, Y.J., 1996, in: Proc. 10th Anniversary HTS Workshop (World Scientific, Singapore) p. 68. Uemura, Y.J., G.M. Luke, B.J. Sternlieb, J.H. Brewer, J.E Carolan, WN. Hardy, R. Kadono, J.R. Kempton, R.E Kiefl, S.R. Kreitzman, P. Mulhem, T.M. Riseman, D.L1. Williams, B.X. Yang, S. Uchida, H. Takagi, J. Gopalakristman, A.W Sleight, M.A. Subramanian, C.L. Chien, M.Z. Cieplak, G. Xiao, VY. Lee, B.W. Statt, C.E. Stronach,

WJ. Kossler and X.H. Yu, 1989, Phys. Rev. Lett. 62, 2317. Uemura, Y.J., L.E Le, G.M. Luke, B.J. Sternlieb, W.D. Wu, J.H. Brewer, T.M. Riseman, C.L. Seaman, M.B. Maple, M. lshikawa, D.G. Hinks, J.D. Jorgensen, G. Saito and H. Yamochi, 1991, Phys. Rev. Lett. 66, 2665. Ulm, E., J.T. Kim, T. Lemberger, S.R. Foltyn and X. Wu, 1995, Phys. Rev. B 51, 9193. Valles Jr, J.M., and R.C. Dynes, 1990, Mater. Res. Soc. Proc. 195, 375. Valles Jr, J.M., A.E. White, K.T. Short, R.C. Dynes, J.P. Garno, A.EJ. Levi, M. At~zlowar and K. Balwin, 1989, Phys. Rev. B 39, 11599. van Ben~rn, P.J.M., A. Wittlin, M.E.J. Boonman, M. Gross, S. Uchida and J. Tamasaku, 1997, Physica C, 293, 136. van der Marel, D., 1996, in: Proc. 10th Anniversary HTS Workshop on Physics, Material and Applications (World Scientific, Singapore) p. 357. van der Marel, D., and J.H. Kim, 1995, J. Phys. Chem. Solids, 56, 1825. van Harlingen, D.J., 1995, Rev. Mod. Plays. 67, 515. Varma, C.M., P.B. Littlewood, S. Schmitt-Rink, E. Abrahams and A.E. Ruckenstein, 1989, Phys. Rev. Lett. 63, 1996. Volovik, K.E., 1993, JETP Lett. 58, 469. Wachter, P., B. Bucher and R. Pittini, 1994, Phys. Rev. B 49, 13164. Walker, M.B., and T. Timusk, 1995, Phys. Rev. B 52, 97. Wang, N.L., S. Tajima, A.I. Rykov and K. Tomimoto, 1998, Phys. Rev. B 57, Rl1081. Warren, W.W, R.E. Walstedt, G.E Brennert, R.J. Cava, R. Tycko, R.E Bell and G. Dabbagh, 1989, Phys. Rev. Lett.62, 1193. Watanabe, T., T. Fujici and A. Matsuda, 1997, Phys. Rev. Lett. 79, 2113. Weber, C., R. Yoshizaki and D.N. Basov, 1997, unpublished. Welp, U., S. Fleshier, WK. Kwok, J. Downey, Y. Fang and G.W Crabta'ee, 1990, Phys. Rev. B 42, 10188. Williams, G.E, R.C. Budhani, C.J. Hirschmugl, G.L. Carr, S. Perkowitz, B. Lou and T.R. Yang, 1990, Phys. Rev. B 41, 4752. Wu, D.H., J. Mao, S.N. Mao, J.L. Peng, X.X. Xi, T. Venkatesan, R.L. Greene and S.M. Anlage, 1993, Phys. Rev. Lett. 70, 85. Xiang, T., and J.M. Wheatley, 1995, Phys. Rev. B 51, 11721.

INFRARED PROPERTIES OF HTSC: EXPERIMENTAL OVERVIEW Xiang, T., and J.M. Wheatley, 1996, Phys. Rev. Lett. 76, 134. Yagil, Y., and E.K.H. Salje, 1996, Physica C 256, 205. Yoshinari, Y., H. Yasuoka, Y. Ueda, K. Koga and K. Kosuge, 1990, J. Phys. Soc. Jpn. 59, 3698. Yoshizaki, R., I. Tomitsuka, K. Ueno and H. Ikeda, 1996, J. Low Temp. Phys. 105, 927. Zaanen, J., G.A. Sawatzky and J.V~ Allen, 1985, Phys. Rev. Let 55, 418.

507

Zha, Y., S.L. Cooper and D. Pines, 1996, Phys. Rev. B 53, 8253. Zhang, K., D.A. Bonn, S. Kamal, R. Liang, D.J. Baar, W.N. Hardy, D.N. Basov and T. Timusk, 1994, Phys. Rev. Lett. 73, 2484. Zhang, S.-C., 1997, Science 275, 1089. Zibold, A., M. Durller, A. Gaymann, H.R Geserich, A. Erb and G. Miiller-Vogt, 1990, Physica C 171, 151.

Handbook on the Physics and Chemistry of Rare Earths VoL 31 edited by K.A. Gschneidnet; Jr., L. Eyring and M.B. Maple © 2001 Elsevier Science B.V. All rights" reserved

Chapter 203 ELECTRONIC AND MAGNETIC R A M A N SCATTERING STUDIES OF THE HIGH-Tc CUPRATES S.L. COOPER

Department o f Physics, Frederick Seitz Materials Research Laboratory, and Science and Technology Center f o r Superconductivity, 1110 W. Green St., University o f Illinois at Urbana-Champaign, Urbana, IL 61801-3080, USA

Comems

1. 2.

3.

4.

List of symbols and abbreviations Introduction Fundamentals of Raman scattering 2.1. Classical description 2.2. Kinematical constraints 2.3. The Raman scattering cross-section 2.3.1. Basic results 2.3.2. Phonon and magnon scattering 2.3.3. Effects of the material's optical response 2.3.4. Resonance effects 2.4. Raman scattering selection rules 2.5. Electronic Raman scattering 2.5.1. Fundamentals 2.5.2. The effective mass approximation 2.5.3. The collision-dominated scattering response Raman scattering in the antiferromagnetic insulator phase 3.1. Two-magnon scattering 3.1.1. Overview 3.1.2. Resonance effects 3.1.3. Influence of doping 3.1.4. Temperature dependence and spin-phonon coupling 3.1.5. Dependence on lattice parameter 3.2. Chiral spin fluctuations 3.3. Excitonic scattering 3.4. Crystal-field excitations Raman scattering in the "normal" metallic phase

4.1. ab-plane electronic scattering in the optimally doped and overdoped regimes 4.2. c-axis electronic scattering 4.3. Raman scattering in the underdoped cuprates 4.3.1. Presence of antiferromagnetic correlations 4.3.2. Evidence for a pseudogap 4.3.3. Evidence for normal-state pairing 5. Electronic scattering in the superconducting state 5.1. Overview 5.2. Theory of Raman scattering in superconductors 5.2.1. The Raman scattering crosssection 5.2.2. Coulomb screening effects 5.2.3. The influence of gap anisotropy 5.3. Analysis of the superconducting response in high-To superconductors 5.3.1. The dx2_),2 pairing state 5.3.2. Effects of orthorhombic distortions 5.3.3. Multilayer effects 5.3.4. Other reported pairing states 5.3.5. Relationship to the "normal"-state response 5.4. Electronic Raman scattering in high magnetic fields 6. Conclusions Acknowledgements References

510 510 513 513 514 514 514 515 515 516 516 518 518 519 521 522 522 522 524 526 528 529 529 530 532 532 509

532 536 538 539 539 541 542 542 544 544 547 547 548 548 549 550 550 553 553 555 555 556

510

S.L. COOPER

List o f s y m b o l s a n d a b b r e v i a t i o n s 219, 3D

A(r) BCS D

E~, E~(r,t), Es(r,t ) H Hin t

Imx J Jj

j(r),jq k~, k s MFL N(0) hi,/~/s n(co) P

P(r,t) p q r r0 S

S(q,co) Tc

Two-dimensional, three-dimensional Vector potential Bardee~Cooper-Schrieffer Diffusion constant Fermi energy Incident and scattered electric field vectors as a function of position and time Magnetic field Interaction Hamiltonian Imaginary part of the Raman response function Superexchange coupling Interlayer exchange coupling Particle current density and qth Fourier component of tile particle current density Incident and scattered wavevectors, respectively Marginal Fermi liquid One-electron density of states at the Fermi level Index of refraction of the incident and scattered media Bose thermal factor Incident laser power Polarization vector as a function of position and time Electron momentaxm Excitation wavevector Electron position Thompson radius of the electron Total spin angular momentum Generalized dynamical structure factor Superconducting critical temperature

TN T* T~, Ts t U V

Vq vv X(r,t) r F, F L F,, 7 A ACT e~, es

e(q,co) /~k #

p(r), pq X Zo

.Z(q, co) co~,Ws cop

N6el temperature Cross-over temperature Transmission coefficient of incident and scattered media Hopping energy On-site Coulomb interaction Scattering volume Coulomb interaction matrix element Fermi velocity Generalized dynamical variable as a function of position and time Electron scattering time Electron scattering rate, and electron scattering rate in Lth scattering channel nth irreducible representation of the space group Electron-photon scattering vertex function Superconducting gap Charge-transfer gap Incident and scattered electric field polarization Dielectric function Imaginary part of the Tsenuto function Effective mass tensor Correlation length Particle number density and qth Fourier component of the particle number density Linear electric susceptibility tensor Static susceptibility tensor RamaJa response function Incident and scattered light frequency, respectively Plasma frequency

1. Introduction T h e r e is little q u e s t i o n that the p h y s i c s a n d c h e m i s t r y o f the high-To cuprates h a v e b e e n a m o n g the p r e e m i n e n t r e s e a r c h topics in c o n d e n s e d - m a t t e r p h y s i c s d u r i n g the p a s t d e c a d e . M u c h o f the initial e x p e r i m e n t a l and t h e o r e t i c a l i n v e s t i g a t i o n o f t h e s e m a t e r i a l s w a s a i m e d at c l a r i f y i n g the m e c h a n i s m r e s p o n s i b l e for high-To s u p e r c o n d u c t i v i t y . Yet, as the

RAMAN SCATTERING STUDIES OF THE HIGH-T c CUPRATES

511

T

AF~ doping (x)

FL ,

Fig. 1. Qualitative phase diagram for the high-T~ cuprates, illustrating the cross-over temperatures T T and T~ described in the text, and the antiferromagnetic insulating (AF), superconducting (SC), and Fermi-liquid (FL) phases (adapted from Emery, Kivelson and Zachar 1997).

richness of the cuprate phase diagram has slowly emerged, a broader goal has evolved to understand the nature of, and relationship among, the many diverse phases associated with these materials. Indeed, it is generally believed that the cuprates exhibit at least four distinct regimes as a function of doping (see fig. 1) (Batlogg and Emery 1996, Emery, Kivelson and Zachar 1997): (i) A n t i f e r r o m a g n e t i c i n s u l a t o r - At zero doping, the cuprates are spin S = ½ antiferromagnetic charge-transfer insulators (TN ~ 300K) with the spins localized primarily on the Cu 2+ ions. From optical measurements, the in-plane Cu dx2 y2 - O Px,y chargetransfer gap in these systems is known to be roughly ACT ~ 1.75 eV (Uchida et al. 1991, Cooper et al. 1993b), while neutron scattering (Aeppli et al. 1989) and two-magnon Raman scattering (Lyons et al. 1988a, 1988b) measurements indicate that the insulating cuprates have a Cu-Cu superexchange coupling constant of J ~ 120-140 meV With increased hole doping in the plane, the 3D N~el temperature decreases rapidly due to a reduction of both the in-plane antiferromagnetic correlation length, ~2D, and the interlayer exchange coupling, J ± (Aharony et al.

1988). (ii) U n d e r d o p e d s u p e r c o n d u c t o r s - The underdoped cuprate superconductors have superconducting Tc's that are suppressed below their optimal values, and are "bad" metals in the sense that the carrier mean free path is less than the de Broglie wavelength (Emery and Kivelson 1995). Two cross-over temperatures have been identified with the underdoped superconductor phase (Batlogg and Emery 1996, Emery, Kivelson and Zachar 1997): (i) a temperature, T~, below which charge inhomogeneities (i.e., "stripes") and short-range antiferromagnetic correlations appear to develop in the CuO2 planes; and (ii) a temperature, T~ (T~ > T~ > Tc), characterized by the development of a "pseudogap". This pseudogap is manifested in a suppression of the spin susceptibility, as inferred from inelastic neutron scattering (Rossat-Mignod et al. 1991) and Cu spin relaxation nuclear magnetic resonance (NMR) measurements (Warren et al. 1989, Walstedt et al. 1990), as well as in a suppression of the electronic density of states, as observed in measurements of

512

S.L. COOPER

specific heat (Ghiron et al. 1992, Loram et al. 1993), dc resistivity (Ito et al. 1993, Batlogg et al. 1994), in-plane and c-axis optical conductivity (Homes et al. 1993, Puchkov et al. 1996), and angle-resolved photoelectron spectroscopy (ARPES) (Loeser et al. 1996, Ding et al. 1996, Marshall et al. 1996). Unlike conventional superconductors, there is evidence that pair formation and phase coherence do not occur simulaneously at Tc in the underdoped cuprates (Micnas et al. 1988, 1990, Batlogg and Emery 1996). Rather, it has been suggested that superconductivity in this phase occurs in a manner akin to a Bose-Einstein condensation; i.e., involving the formation of normal state pairs above a cross-over temperature T*, followed by the onset of phase coherence among the pairs below a lower, "superconducting" transition at To. (iii) O p t i m a l l y d o p e d s u p e r c o n d u c t o r s - At "optimal" doping, the transition temperatures of high-Tc superconductors attain their maximum values, and the two crossover temperatures described above merge near the superconducting transition; consequently, there is an abrupt transition between the normal and superconducting states in the optimally doped regime. With some notable exceptions (Sun et al. 1994, Chaudhari and Lin 1994, Kouznetsov et al. 1997), measurements of the superconducting state in the hole-doped cuprates are largely consistent with a dx2 y~ pairing state (Annett et al. 1996), including phase-sensitive SQUID experiments (Wollman et al. 1993, Brawner and Ott 1994, Van Harlingen 1995), phase-sensitive studies of the Josephson effect in oriented crystal composites (Tsuei et al. 1994), ARPES experiments (Shen et al. 1993, Ding et al. 1994), and Raman scattering measurements (Devereaux et al. 1994a, Devereaux and Einzel 1996a). (iv) O v e r d o p e d s u p e r c o n d u c t o r s - At still higher doping, Tc decreases below its optimaldoping value, and many of the transport and optical properties exhibit features suggestive of Fermi-liquid behavior (Uchida et al. 1991, Batlogg et al. 1994). For example, in this phase the resistivity exhibits a Fermi liquid-like temperature dependence, p ( T ) ~ T 2 (Batlogg et al. 1994). In many respects, Raman scattering is a natural probe of the cuprates. As a technique that is sensitive to spin, electronic, and phonon degrees of freedom, Raman scattering can provide energy, symmetry, and lifetime information about many of the diverse excitations present throughout the complex phase diagram of the cuprates. Moreover, Raman scattering provides an ideal method with which to explore the evolution of charge, spin, and lattice excitations b e t w e e n different phases. Raman scattering lends itself particularly to the study of such important issues as the nature of the superconducting pairing mechanism, the manner in which antiferromagnetic correlations evolve into the metallic phase, and the relationship between "pseudogap" and superconducting phases in underdoped cuprates. The purpose of this chapter is to review Raman scattering studies of the many diverse phases of high-Tc superconductors. As there are already comprehensive review articles dealing extensively with Raman-active phonons and phonon self-energy effects in high-To superconductors (Thomsen and Cardona 1989, Thomsen 1991, Cardona 1999), this review will primarily focus on the nature of electronic and magnetic excitations in the different

RAMAN SCATTERINGSTUDIESOF THE HIGH-Tc CUPRATES

513

phase regimes of the high-Tc superconductors. The outline of this review is as follows: Section 2 discusses basic details regarding the Raman scattering process and technique. Section 3 examines the Raman scattering response in the insulating phase, focusing on scattering from two-magnon excitations, topological ("chiral") spin excitations, excitons, and crystal-field excitations. Section 4 describes the Raman scattering response of the unconventional metallic phase, including the anomalous ab-plane electronic scattering "continuum", interlayer electronic scattering, and electronic scattering in the underdoped phase. Section 5 discusses electronic Raman scattering studies of the superconducting state. Finally, sect. 6 concludes with remarks about possible future directions of Raman scattering in the high-T~ cuprates.

2. Fundamentals of R a m a n scattering 2.1. Classical description

Much of the essential physics of inelastic light scattering can be understood classically by considering the macroscopic polarization P(r,t) induced in the material by an incident electric field E~(r, t), (1)

P(r, t) = xE~(r, t),

where Z is a second-rank tensor describing the linear electric susceptibility of the material. Elastic contributions to the light-scattering response arise from the induced polarization associated with the static susceptibility, X0. On the other hand, inelastic contributions to the light-scattering spectrum arise due to the presence of excited states that dynamically modulate the wavefunctions and energy levels in the material. In the semiclassical description, these excited states are represented by a dynamical variable, X(r,t), which can be associated, for example, with deviations of atomic position, charge density, or spin density from equilibrium. The resulting induced polarization in the presence of excited states in the material is obtained by expanding the dynamical susceptibility X in powers of X around the static susceptibility, X0:

dZ°Xe + ! d2X°x2

P ( t ) = XoE + ~

2 dX 2

-- ±

...

(2)

Thus, in addition to the elastic scattering contribution in eq. (2), inelastic contributions to the scattering response arise from the modulation of the polarization at the characteristic frequency of excitations in the material, X = X0ei(C°t+ q)). In particular, the contribution to the induced polarization that is linear in X is given by p(1) = eei[(~)i ± ~0t+ cp], where col is the frequency of the incident electric field. The resulting inelastic light-scattering cross-section includes two contributions: (a) the Stokes component, in which the incident light field Er(r,t) creates an excitation of energy and wavevector, co and q, causing a

514

S.L. COOPER

reduction in energy of the scattered field Es(r,t), Cos = Col co; and (b) the anti-Stokes component, wherein the scattered field gains energy from thermally excited excitations in the sample, Cos = Col+ Co. In scattering processes having time-reversal symmetry (i.e., non-magnetic excitations) and no resonant enhancement, the principle of detailed balance requires that the Stokes and anti-Stokes intensifies be related by (Hayes and Loudon 1978) /AS(Co)=/S(Co)(CoAS/CoS)2e h~/Ic~T, where Cos (CoAs) is the absolute frequency of Stokes (anti-Stokes) light, Cois the frequency of the excitation, and T is the temperature. 2.2. Kinematical constraints In a pure crystal, i.e., one exhibiting translational invariance, kinematical constraints imposed by wavevector conservation dictate one of the most important constraints on the Stokes Raman scattering process, namely, q : k1 - k s .

(3)

In typical experimental situations one has co<
(4)

As inelastic light scattering experiments are usually performed with incident frequencies in the visible frequency range, CoI ~ 1014 Hz, the range of excitation wavevectors typically probed in light-scattering experiments is 0 < l q I < 3 x 1 0 3 ~ 1 (Hayes and Loudon 1978), which is generally several orders of magnitude smaller than the size of the Brillouin zone boundary, IkzB] ~ 2~c/a ~ 1 A 1 Thus, Raman scattering generally probes only excitations very near the Brillouin zone center; i.e., Iq] ,-~0. This limitation is circumvented under certain circumstances, including when wavevector conservation is lost due to the presence of disorder, and when two-particle (e.g., two-phonon and twomagnon) excitations of equal and opposite momenta, q and -q, are excited. 2.3. The Raman scattering cross-section 2.3.1. Basic results In the quantum-mechanical derivation of the Raman scattering cross-section, one considers the two-photon process in which a photon changes its frequency co, wavevector k, and polarization e via scattering from the values (col, kt, El) to the values (Cos, ks, es), while the scattering material experiences a transition from ]i) to I f ) . The interaction Hamiltonian between light and matter is given by

e/

Hint = --c

d3rj(r) •A(rj) + 2 m ~

d3rp(r)A2(rJ)'

(5)

where A(r) is the vector potential associated with the incident radiation field, j(r) = lzm }-~j (pj6(r -- rj) + 6(r - rj)pj) is the particle current density, pj and r / a r e the momentum and position o f t h e j t h particle, and p(r) = ~ j 6(r - r2) is the particle number density.

R A M A N SCATTERING STUDIES OF THE HIGH-To CUPRATES

515

As Raman scattering is a two-photon process, the scattering cross-section is obtained by taking the A 2 term in eq. (5) to first order in perturbation theory, and the A term in eq. (5) to second order in perturbation theory. The measured quantity in a typical Raman scattering measurement is related to the photon differential cross-section per unit scattered frequency range, d2a/dr2 do)s, which reflects the fraction of incident photons inelastically scattered in the energy interval, dco s, per unit solid angle, dO. The resulting quantum-mechanical photon scattering crosssection is given by d2a _ V2co, co3 ns ~ C4 nI d~2 dcos f

((l(f [X~s[i)l 2) 6(c0~- COs- co),

(6)

where ((...}) represents a thermodynamic average over the initial states of the scattering material. The transition susceptibility tensor X in eq. (6) is the quantum-mechanical analogue of the classical susceptibility in eq. (1), and has matrix elements

(f Ix,sl i)

e2

y,

[ ( f ]es "JksI m) (m I~I 'j-k~ l i)

Vcos~, "-7" L

[e.i --£• --~,~ + ( f ]~i -ik~] m) (m les 'JksI i) 1 [Era - Ei + ho)s] J

e2

- -

mVcoscoj

EI. es ( f I P-ql i ) ,

(7) where .]q = ~ ~ j e - i q ' r j ( p j -- hq/2) is the qth Fourier component of the particle current density operator, and p q = ~ j e-Iqrj is the qth Fourier component of the particle density operator. Thus, the Raman scattering cross-section for a particular excitation is given by the correlation function of the dynamical susceptibility at the wavevector q and frequency co of the excitation.

2.3.2. Phonon and magnon scattering In order to consider light scattering from phonons or spins, one typically treats the electron-phonon or electron-spin interaction as a perturbation and calculates the relevant scattering cross-sections using third- or higher-order perturbation theory (Hayes and Loudon 1978).

2.3.3. Effects" of the material's optical response The dependence of the scattering cross-section in eq. (6) on the optical constants of the incident and scattered media is deceptively benign (~nJns), and it should be noted that this factor is derived by considering only the radiated intensity inside the scattering medium. As the scattering intensity is measured outside the material, the differential scattering

516

S.L. COOPER

cross-section is actually related to the measured scattering intensity, I, by several factors which depend upon the optical response of the material (Reznik et al. 1993),

do d$2 dca

I P TI Ts V ~in'

(8)

where P is the incident laser power, TI and Ts are the transmission coefficients at the air-sample interface for the incident and scattered photons, respectively; V = (ai + as) 1 is the scattering volume, which depends upon the absorption coefficients at the incident (al) and scattered (as) photon frequencies; ~in = $'2out/n2 is the solid angle inside the material collected by the collection optics, where n is the index of refraction of the material, and g2out is the solid angle outside the material depending upon the f-number of the collecting lens. These optical corrections must be carefully accounted for, particularly when comparing Raman spectra as a function of doping. 2.3.4. Resonance effects It is evident from eqs. (6) and (7) that second- and higher-order terms in the scattering cross-section involve transitions to virtual intermediate states that contribute energy denominators of the form ( E m - E i - h o ~ ) and (Em-Ei+ha~s). As a result, "resonant enhancements" of the scattering cross-section occur when the incident photon energy hcol approaches the energy difference between initial and intermediate states, causing the energy denominator to vanish. Various studies of the resonant Raman scattering intensity of phonon and bimagnon excitations as the incident photon frequency is tuned through real electronic transitions have been particularly useful for exploring the coupling between these excitations and the electronic bands in the cuprates; see, for example, Heyen et al. (1990, 1992), Wake et al. (1991), Yoshida et al. (1992), and Blumberg et al. (1996). 2.4. Raman scattering selection rules The selection rules for the Raman scattering process are imposed by the susceptibility tensor in eqs. (1) and (6). Consider the matrix element (ftx~b[i), where [i) and [f) are the initial and final states of the scattering medium, which transform like the irreducible representations of the material's point group, Fi and Ff, respectively, and where the susceptibility tensor X associated with an excitation is decomposed into irreducible representations Fn of the point group of the crystal, Z,b = ~

anXr,,.

(9)

tt

The transition susceptibility matrix element (f'[Xabti) is nonzero only if the decomposition of the direct product Pj* ® Fn ® Fi contains the totally symmetric representation, /~1~ or equivalently, if the decomposition of F~ ® Fi contains F,,. Additionally, for an excitation to be "Raman-active', the relationship between the induced polarization P, the susceptibility

RAMAN SCATTERING STUDIES OF THE HIGH-T~ CUPRATES

517

Table 1 Irreducible representations of the allowed Raman tcnsors in both orthorhombic (D2h) and tetragonal (D4h) point groups Point group

Alg

Orthorhombic (l)2h)

Tetragonal (D4h)

xx YY zz )

xx YY zz )

A2~

.

(,x )

B2g

(zx xZ)

E~

(zx xz)( zxXZ)

tensor X, and the incident field El in eq. (1) must be covariant under all spatial transformations of the scattering medium. Since P~ and El are both polar vectors, this is equivalent to stating that an excitation is Raman-active only if it has a symmetry contained in the decomposition of F~v ® Fpv. Table 1 lists the specific allowed Raman tensors for the orthorhombic Dzh and tetragonal D4h point groups to which most high-To cuprates belong. An extremely powerful feature of the Raman scattering technique, implicit in the susceptibility tensor coefficients in eq. (7), is that one can experimentally couple to different components of the susceptibility tensor - and thereby identify the symmetry of a given excitation - by varying the polarizations of the incident and scattered light relative to the sample's crystalline axes. The nomenclature typically used for describing the geometry of a Raman scattering experiment is kl(ei es)ks, where ki and ks are the wavevector directions of the incident and scattered photons, respectively, and e1 and es are the polarization directions of the incident and scattered photons, respectively. For example, z(x,y) ~ indicates a depolarized, "true-backscattering" geometry in which the incident and

518

S.L. COOPER

scattered wavevectors are directed along the z- and -z-axes of the crystal, respectively, and the incident and scattered photons are polarized along the x- and y-axes, respectively. 2.5.

Electronic Raman scattering

2.5.1. Fundamentals As this review is primarily concerned with low-energy electronic scattering in the "normal" and superconducting states o f high-Tc superconductors, it is useful to consider in detail the cross-section for Raman scattering from intraband electronic excitations. Using eqs. (6) and (7), the electronic Raman scattering cross-section can be related to a correlation function associated with an effective density p, d 2o _ r 2 cos ns Z dr2 dcos o col Hi f

(([(f [Pq t i)[ 2)) ?~(Eu-

E i -

(1 O)

hco),

where ((...)) represents a thermal average over the initial states i, Pq

:

Z

~nkCnk+q,
(11)

n, k, (~

c t and c are the electron creation and almihilation operators, respectively, r0 = eZ/mc2 is the Thomson radius of the electron, n is a band index, and 7nk is the electron-photon Raman scattering "vertex". In the case of intraband electronic excitations, the initial and final band states in eq. (7) are taken to be the same, and 7nk is given by (Abrikosov and Genkin 1974)

7nk

EI

m

ES +

1 ~

[(nktes

m~

L

"pln'k+ki)(n'k+ki le~ .pink) Enk -En'k+k~ + hcot

(nk le~ "p[ n ' k +

ks) (n'k- ks I,~s 'Pl nk) ]

Enk -- E n ' k -

ks -- h c o S

(12)

J '

where the electronic states are represented by their band indices, n and n', and wavevectors, k, and it has been assumed that Ik~l ~ lksl (see sect. 2.2). The cross-section in eq. (10) can be equivalently written d2a ~ dg2 dcos - r°2

~S(q, co),

(13)

where S(q, co) is a generalized dynamical structure factor, which can be related to the imaginary part of a Raman response function X~ via the fluctuation-dissipation theorem (Hayes and Loudon 1978), 1

S(q, co) = - ~ [1 + n(co)] Imz~(q, co),

(14)

where n(co) = [exp(hco/kBT) - 1] -j is the Bose thermal factor. )(~ is the Raman response function associated with fluctuations in the effective density ~, and can be evaluated

RAMAN SCATTERING STUDIES OF THE HIGH-To CUPRATES

519

to various levels of approximation using Green's function techniques. In particular, various self-energy "vertex" corrections are often included in Z ~ ( q , o)) when considering electronic Raman scattering in metals and superconductors (see sect. 5.2.1), including those associated with electron electron interactions (Virosztek and Ruvalds 1991, 1992), electron-impurity scattering (Zawadowski and Cardona 1990), electron-phonon scattering (Itai 1992), and Coulomb screening (Klein and Dierker 1984). 2.5.2. The effectioe mass approximation Importantly, the Raman scattering vertex ]/~k in eq. (12) can be directly related to the inverse effective mass tensor W l, m

]/,k = el" -- " ES = m ~

~1

i,j

02Enk ESih20k i Okj I?'lj'

(15)

under certain circumstances, namely: (a) when the incident and scattered photon energies are far from a resonance condition, he)l, h~os << ]E, - E , , 1, so that the photon energies h~o~, hCOs in the denominators of eq. (12) can be ignored; and (b) when virtual intraband transitions can be ignored, so that the sum in eq. (12) can be restricted to bands n' ~ n (Strohm and Cardona 1997a). Within this "effective mass approximation", the electronic Raman scattering cross section in eq. (10) can be associated with effective mass fluctuations around the Fermi surface, the different symmetry components of which can be experimentally selected by varying the scattering geometry (i.e., e~ and Es) in the Raman scattering experiment (see eq. 15). It is often useful to explicitly account for the crystalline symmetry of the material in the Raman scattering cross-section by expanding the Raman scattering vertex ]/,k in terms of Fermi surface harmonics q~L (Allen 1976, Klein and Dierker 1984, Devereaux and Einzel 1995),

]/k = ~

(16)

]/~t~tL'(k), L

where L refers to the Lth-order contribution to the scattering vertex transforming like the ~th irreducible representation of the crystal's point group. In the case of the tetragonal D4h point group of the cuprates, the lowest-order terms for the vertices are given by (Devereaux and Einzel 1995),

] /= ]/0A~- ]/Alg l gCOS 4q0,

I/Big k = YB,~cos2cp,

]/B2g k

]/B2gsin2cp.

(17)

Equation (17) indicates that the Blg and B2g vertex contributions preferentially select Fermi surface regions (kx, ky) near { ( ±sv/a, 0), (0, -t-~/a) } and (+Jr/a, -4-~/a), respectively, while the Alg vertex samples more uniformly around the Fermi surface (see fig. 2). Note that the isotropic part (L = 0) of the Raman scattering vertex in eqs. (15) and (16), which comprises the entire electronic scattering response in the absence of interband

520

S.L. COOPER ky

g2g kx

C

Fig. 2. The functionaldependenceof the Raman scattering vertices, Y,k, associated with the tetragonal D4h point group of the cuprates for: BI,~(7~lg) symmetry;B2g (722g)symmetry;and Alg (]/kA 1 g) symmetry,for the case 7Alg=0 (see eq. 17). transitions or Fermi surface anisotropy, corresponds to scattering from ordinary density fluctuations, and has a Raman scattering cross-section given by d2a, df2 d o s

r2

hq 2V F -1 7 (o)s) 2 ~-~ (el" es) 2 [n(o))+ 1] A ~ g ~ 2 1 m / - - / 4ere

Le(q, o)) J '

(18)

where Im[-1/e(q,o))] is the dielectric loss function. The dynamical structure factor S(q,o)) associated with density fluctuations satisfies the f-sum rule (Pines 1963), f~'~ do) o)S(q, o)) = Nq2/2m, where N is the number of particles of mass m. In the absence of strong screening by the long-range Coulomb interaction, most of this spectral response is associated with single-particle excitations, which are constrained by energy and crystal momentum conservation to fall in the energy range 0
RAMAN SCATTERING STUDIES OF THE HIGH-To CUPRATES

521

are not screened by the long-range Coulomb interaction, and do not satisfy a sum rule in general - consequently, the Raman scattering response associated with L ~ 0 mass fluctuations can have a large integrated intensity. In addition to anisotropic contributions to the Raman scattering vertex, Kosztin and Zawadowski (1991) have pointed out that intense intraband electronic Raman scattering in metals requires either impurity scattering, inelastic scattering, or strong electron-electron interactions to take up the momentum in the electronic scattering process between different regions of the Fermi surface. As a result, the L ~ 0 Raman scattering response is not limited by momentum considerations to frequencies less than quv. 2.5.3. The collision-dominated scattering response Ipatova et al. (1981) first pointed out, in the context of "intervalley" electronic Raman scattering in doped semiconductors such as Ge and Si (Chandrasekhar et al. 1977, Contreras et al. 1985), that the electronic mean free path is often shorter than the optical penetration depth in the presence of strong electronic scattering; i.e., l < 6 = q-i. In this case, the intraband electronic Raman scattering cross section is more appropriately described by a "collision-dominated" response (Ipatova et al. 1981, Zawadowski and Cardona 1990),

~orL SL -- [n((,O) -k 1] I~L[2 (D2 q - F / ? '

(19)

where SL(~O) is the electronic scattering response in channel L, FL is the carrier scattering rate in channel L, and the amplitude f a c t o r I]/LI2 is the square of the Raman scattering vertex. Note that this Raman spectral response is quite similar to the Drude optical response, with the exception that the prefactor in the latter is a scalar quantity related to the squared plasma frequency, while the prefactor in eq. (19) is a symmetry-dependent factor related to the effective mass tensor. Therefore, by measuring the intraband Raman scattering response in different scattering geometries, one can isolate the carrier scattering rates associated with different scattering "channels", L (Zawadowski and Cardona 1990). On the other hand, unlike the optical conductivity, there is no sum rule on the intraband Raman scattering response. In addition to electronic scattering in doped semiconductors, the collision-dominated form has been used to describe electronic scattering in the presence of disorder (Zawadowski and Cardona 1990); strong electron-phonon scattering (Itai 1992); and electron-electron correlations in the cuprates (Klein et al. 1989, Cooper and Klein 1990), titanates (Katsufuji and Tokura 1994), manganites (Yoon et al. 1998), and hexaborides (Nyhus et al. 1997). In the absence of strong electron-electron correlations or inelastic scattering, the carrier scattering rate in eq. (19) is generally taken to be frequency-independent, F = 1/T +Dq 2, where T is the carrier scattering time and D is the diffusion constant (Ipatova et al. 1981, Chandrasekhar et al. 1977, Contreras et al. 1985). However, for strongly correlated systems, the scattering response in eq. (19) can be "extended" to account for correlation effects by using a frequency-dependent scattering rate, F=-F(co, T) in eq. (19), in a

522

S.L. COOPER

manner analogous to that done in the "extended Drude model" for the optical conductivity of strongly correlated systems (Uchida et al. 1991, Cooper et al. 1993b). For example, an "extended" collision-dominated scattering model has been used to describe the electronic Raman scattering responses of SrxLal xTiO3 (Katsufuji and Tokura 1994) and EuB6 (Nyhus et al. 1997), employing eq. (19) and a "Fermi-liquid" scattering rate given by F(co, T) = Fo(T) + ao) 2, as well as the electronic Raman scattering response of the cuprates, which has been commonly described using eq. (19) with a scattering rate given by the Marginal Fermi Liquid (MFL) form, F(co, T) = aV/~O2 +/32T 2 (Klein et al. 1989, Cooper and Klein 1990).

3. Raman scattering in the antiferromagnetic insulator phase It is now well known that in the insulating phase, high-Tc cuprates are spin S = ½ antiferromagnetic charge-transfer insulators (TN ~ 300 K) with the spins localized primarily on the Cu 2+ ions, and with a superexchange interaction between spins of roughly J ~ 120 meV. A number of interesting excitations have been observed by Raman scattering in this phase, including two-spin excitations (Lyons et al. 1988a,b), "chiral" spin fluctuations (Shastry and Shraiman 1990, 1991, Khveshchenko and Wiegmann 1994, Sulewski et al. 1991), excitonic excitations (Liu et al. 1993a,b, Salamon et al. 1995), and crystal-field transitions associated with the lanthanide ions (Heyen et al. 1991, Rufet al. 1992, Dufour et al. 1995, Sanjurjo et al. 1995, Jandl et al. 1995). These excitations are considered in greater detail in the following. 3.1. Two-magnon scattering 3.1.1. Overview Raman scattering studies of the antiferromagnetic insulating cuprates have primarily focussed on two-magnon scattering, which arises when incident photons excite two magnons of opposing momenta, q and -q. Much of the basic theory of two-magnon scattering evolved from attempts to understand Raman scattering measurements of "conventional" antiferromagnets such as K2NiF4 [S= 1] and MnF2 [S = ~] (Fleury and Loudon 1968, Fleury and Guggenheim 1970) (see fig. 3). The two-magnon scattering process occurs in three steps (see fig. 6), including electric dipole transitions in the initial and final steps and a double spin-flip in the intermediate step, and is described by the Hamiltonian (Parkinson 1969)

M R : ~ (El.~if)(Es.cTij)Si.Sj, (20) 0 where El and Es are the incident and scattered electric fields, respectively, a!/ is a unit vector connecting sites i and j, and S is the spin operator. In the absence of anisotropy and magnon-magnon interaction effects, the zero-temperature magnon energy in a 2D Heisenberg antiferromagnet is given by E 2 = (SJZ) 2 [1 - 1 (cos(kxa) + cos(kya))],

(21)

where Z (= 4) is the number of nearest neighbors, S is the spin, and J is the exchange interaction constant. As two-magnon Raman scattering is dominated by short-wavelength

RAMAN SCATTERINGSTUDIESOF THE HIGH-Tc CUPRATES X(YX|Y

523

NI0°K

t t,-

! '

.*.,

.d

\

/ I

..i- -.7.- ~ 400

450

.5oo

550

600

(cm-') (b) XY INSTRUMENTAL WIDTH

..t ¢¢.

tn

P, z i-

_z

SO 85 9 0 9 5 ~ 0 IO5 IIO FREQUENCY SHIFT (¢m-'~

Fig. 3. (top) Two-magnonRaman scattering response of the spin-1 antiferromagnetK2NiF4 (after Fleury and Guggenheim 1970), and (bottom) two-magnon scattering from the spin-5/2 antiferromagnet MnF2 (after Fleuryand Loudon 1968).

spin excitations near the zone boundary where the density of states is large, the peak in the two-magnon scattering response of the spin-½ cuprates is predicted by eq. (21) to be hcoo = 2 S J Z = 4 J . However, magnon-magnon interaction effects are expected to reduce the energy of the two-magnon scattering peak by roughly 2 J S (Parkinson 1969, Lyons et al. 1988b), resulting in a two-magnon energy of h~o0 = 3J. A more precise accounting of the two-magnon scattering response using linear spin-wave theory gives a two-magnon peak energy at ho)0=2.7J for spin-½ antiferromagnets (Elliot et al. 1968, Elliot and Thorpe 1969, Parkinson 1969). Notably, for spin interactions involving only nearestneighbor spins, spin-½ antiferromagnets with D4h symmetry are predicted by eq. (20) to have two-magnon Raman scattering intensity only in the Big scattering geometry. In the insulating cuprates, strong two-magnon scattering has indeed been observed in the Big geometries of La2CuO4 (Lyons et al. 1988a), YBa2Cu306 (Lyons et al. 1988b), the T~-phase compounds such as Nd2CuO4 and Sm2CuO4 (Sulewski et al. 1990, Tokura et al. 1990, Cooper et al. 1990), and Sr2CuO2C12 (Blumberg et al. 1996), from which exchange parameters of J = 136 meV, J = 117 meV, J = 110 meV, and J = 130 meV, respectively, were deduced. However, the observed two-magnon response in the cuprates differs in several important respects from the predictions of the standard approximations of eq. (20). First, in addition to strong Big-symmetry scattering, weak two-magnon scattering is also

524

S.L. COOPER I

/~.

I

!

S=O

I

I

8 • 0.6 TC =60K

(a)

I

(hi i~, I

I

I

(¢)

8 , tO

~ T ¢ =¢JOK

Q. U

I-Zlal

tz

O

2000

6000

I

I o 2ooo 60oo ENERGY SHIFT (cm'l)

Bltg 1

2000

!

I

6000

Fig. 4. Alg- , Big- , and B2g-symmetry roorn temperature Raman scattering spectra of YBa2Cu3Or+ o in (a) the antiferromagnetic insulating phase (6=0), (b) the underdoped superconductor phase (6=0), and (c) the optimally doped superconductor phase (6 = 1) (after Lyons et al. 1991).

observed in the A l g and B2g geometries (Singh et al. 1989, Sulewski et al. 1990, Sugai et al. 1990a,b) (see fig. 4). This has been attributed in part to the importance of next-nearestneighbor and longer-range spin interactions in eq. (20) (Singh et al. 1989). Second, the two-magnon lineshape observed in the insulating cuprates is significantly broader than that observed in spin-1 antiferromagnets such as KzNiF4, and is also extremely asymmetric, with a high-energy tail extending far beyond the classical cutoff energy of 4J. Singh et al. (1989) attributed part of this broadening to quantum fluctuations associated with the small spin (1) of the cuprates. Multi-magnon processes are likely to be primarily responsible for the high-energy tail of the two-magnon response in the cuprates (Sugai et al. 1990a,b, Rtibhausen et al. 1996), although calculations by Canali and Girvin (1992) suggest that the four-magnon intensity may be too weak to contribute a substantial peak near 4J Other contributions to the broad, asymmetric two-magnon scattering response of the cuprates have been suggested (Bacci and Gagliano 1991, Haas et al. 1994, Prelovsek and Jaklic 1996, Sandvik et al. 1998), including spin-phonon interactions (Weber and Ford 1989, Nori et al. 1992, 1995, Lee and Min 1996), and hopping disorder (Basu and Singh 1997). In particular, Freitas and Singh (2000) recently proposed that many of the discrepancies between the predictions of the Fleury-Loudon-Elliot theory of two-magnon scattering and the observed two-magnon spectra of the cuprates can be accounted for by considering spin-phonon interaction effects (Lorenzana and Sawatzky 1995a,b); these cause phonons to act as momentum sinks in the two-magnon scattering process, thereby allowing twomagnon states with arbitrary total momentum to contribute to the scattering response. 3.1.2. Resonance effects Studies of the two-magnon spectrum as a function of incident photon energy in YBa2Cu306 and Sr2CuO2C12 (Lyons et al. 1988b, Sugai 1989, Yoshida et al. 1992,

RAMAN SCATTERING STUDIES OF THE HIGH-To CUPRATES 2.8J + J . ~ \ /_\4J

/:

YBa2Cu3On.1

+JL

x'y'

i xo.1

/1" "~

+o

\\

._

525

+

+

, + , ,, . . ,

~

•

•

• 1.92 eV (2~+1.4J)

3000

4000

i

1000

2000

.

.

.

.

5000

Raman Shift (cm "+)

& .._~ C

=m t::: m ¢

3

/

~.l)lre'l

O,o

0

0~

•

•

YBa2Cu3Oe..~ , 1.5

-,

~O

i 2 m~ (eV) 2.5

3

Fig. 5. (top) B+g-symmetry two-magnon Raman scattering cross section in single crystals of YBaaCu306.~ for different excitations above the charge-transfer gap 2A [=ACT]. (bottom) Intensity of the Big two-magnon peak at 2.8J as a function of excitation energy c% Also shown for comparison is the imaginary part of the dielectric function, e2 (after Blumberg et al. 1996).

C o o p e r et al. 1993b, Liu et al. 1993b, B l u m b e r g et al. 1996) demonstrate that twom a g n o n scattering is dramatically e n h a n c e d when incident photons are at least 2 J 3 J above the charge transfer gap energy, AcT (see fig. 5). Qualitatively, this behavior is consistent with the specific three-step process responsible for t w o - m a g n o n scattering in the cuprates (see fig. 6): (i) an i n c o m i n g p h o t o n excites a spin-up Cu(3d 9) hole to an

526

S.L. COOPER

(a) ~-

~

(b) 0 site

~

--iL" 0

ACT

•--l-

-t-

Cu site i

ou

Cu site j (d)

(c)

Cu

-t-ou

Cu

Cu

+

Cu

Fig. 6. Diagram of the two-magnon Raman scattering process in the cuprates: (a) initial antiferromagnetic spin configuration on the Cu sites; (b) incident photon excitation of a spin-up Cu(3d 9) hole to an intermediate-state O(2p) orbital; (c) double spin flip due to direct spin exchange between the spin-up hole on the intermediate-state O(2p) site and a spin-down hole on a nearest-neighbor Cu site; (d) hopping of a spin-down hole from the O(2p) intermediate state to the original copper site, emitting the scattered photon.

intermediate-state O(2p) orbital (fig. 6b), a process which requires photon energies in excess of the charge-transfer gap, AcT; (ii) a double spin-flip results from direct spin exchange between the spin-up hole on the intermediate-state O(2p) site and a spindown hole on a nearest-neighbor Cu site (fig. 6c); (iii) the newly created spin-down hole on the O(2p) intermediate state returns to the original copper site (fig. 6d), emitting a photon. Several recent calculations using a spin-density wave formalism for the halffilled Hubbard model support this picture. In particular, these calculations suggest that the two-magnon resonance behavior in the cuprates is governed by a "triple resonance" condition, which arises from the vanishing of the three energy denominators in the twomagnon Raman scattering cross section associated with the three processes in fig. 6 [i.e., steps (i), (ii) and (iii) above] (Chubukov and Frenkel 1995, Morr and Chubukov 1997, Sch6nfeld et al. 1997). This "triple resonance" behavior accounts for several significant features observed in YBa2Cu306.t, Sr2CuO2C12 (Blumberg et al. 1996), and PrBa2Cu307 (Rtibhausen et al. 1996), including: (a) a linear dependence of the inverse scattering intensity on excitation frequency above the charge transfer gap; (b) the appearance of two peaks in the two-magnon response, one at the maximum two-magnon energy, 4J, and another at roughly 2.8J due to magnon-magnon interactions; and (c) the presence of two-magnon scattering intensities in the Alg and Big geometries. 3.1.3. Influence of doping With increased hole doping, the two-magnon Raman intensity of the cuprates rapidly diminishes in intensity and shifts slightly to lower frequencies (see figs. 4 and 7), behavior generally attributed to spin disordering of the system by the doped carriers (Lyons et al. 1988a,b, 1991, Reznik et al. 1992, Tajima et al. 1993, Blumberg et al. 1994, Rfibhausen et al. 1997, 1999). Because the two-magnon scattering process in the cuprates is resonant with the charge-transfer band, the dhninution of scattering intensity

RAMAN SCATTERINGSTUDIESOF THE HIGH-Tc CUPRATES

I T=14K

:

I1

0.0

~>,

0.3

.~_

0.7,

0

0

,

Y: I I

.~

527

°ping(Y'cOntent)

2500 Raman shift (cm -1)

5000

Fig. 7. Doping dependenceof the Big Raman spectra of Yl_~Pr,Ba2Cug_vAlyO7 crystals at T=14K. The solid linerepresentsthe doping-dependentpeakposition of the two-magnon response (after Riibhausen et al. 1997).

with doping may also be affected by changes in the Raman scattering vertex due to the deleterious effects of doping on the charge-transfer band (Cooper et al. 1993b). There is some controversy regarding the persistence of two-magnon scattering into the metallic/superconductor phase with increased doping. Early reports by Krol et al. (1989) and Reznik et al. (1992) found that two-magnon scattering disappears in the superconducting phases of YBa2Cu306~x. However, Lyons et al. (1991), and more recently, Blumberg et al. (1994), and Rtibhausen et al. (1997, 1999), reported that twomagnon scattering persists well into the metallic phases of YBa2Cu4Q, YBa2Cu306+x, Y1 xPrxBa2Cu3OT, and Bi2Sr2CaCu2Os, suggesting that antiferromagnetic fluctuations with a spatial extent of several lattice constants survive into the underdoped metallic phase (see figs. 4 and 7). A conclusive experimental resolution of this controversy is complicated by the fact that some of these materials are prone to a macroscopic mixing of phases away from optimal doping (Iliev et al. 1993). However, other measurements such as neutron scattering (Hayden et al. 1996) and nuclear magnetic resonance (NMR) (Pennington et al. 1989) have also found that overdamped spin excitations are present in the metallic phase of the cuprates; hence it is reasonable to expect that two-magnon scattering should also be present. Moreover, two-magnon excitations are dominated by short-wavelength spin excitations at the zone boundary that should be rather resilient to doping. Two-magnon scattering was also found to persist well into the metallic phase of the electron-doped system Ndz_xCexCuO4 (Sugai and Hidaka 1991, Tomeno et al. 1991), offering further evidence for the robusmess of short-range antiferromagnetic correlations against doping

528

S.L. COOPER

in the cuprates, hnportantly, two-magnon scattering seems to disappear upon doping into the overdoped phase of Bi2Sr2CaCu208 (Rfibhausen et al. 1999), suggesting that the antiferromagnetic correlation length is less than twice the lattice parameter in the overdoped phase regime.

3.1.4. Temperature dependence and spin phonon coupling Knoll et al. (1990) studied the temperature dependence of the Big two-magnon response in YBa2Cu306 (R=Y, Eu) (see fig. 8), finding a much more dramatic increase in the two-magnon linewidth than can be accounted for by damping due to magnon-magnon scattering or spin diffusion (Brenig and Monien 1992, Kampf and Brenig 1992). These authors consequently argued that spin-phonon scattering is the dominant relaxation mechanism for spin excitations in the insulating cuprates. Nori et al. (1995) further suggested, based on calculations using exact diagonalization and quantum Monte Carlo techniques, that the spin-phonon interaction is responsible not only for the broad, asymmetric lineshape of the two-magnon scattering response, but also for the presence of an Alg-symmetry two-magnon response. The importance of the spin-phonon interaction to spin excitations in the insulating cuprates was also stressed in a slightly different context by Lorenzana and Sawatzky (1995a,b), who argued that sharp absorption peaks observed near 0.35 eV in the optical spectra of several insulating cuprates (Perkins et al. 1993) are actually two-magnon excitations that are infrared-allowed due to weak coupling with an optical phonon. Indeed, from fits to the infrared spectra with their model, Lorenzana and Sawatzky infer values for J comparable to those estimated from twomagnon Raman scattering measurements. 50

T=11K

40 e~

5

3O

f T = 320 K

2o

.X/~. T= 470 K

10

0

2000 4000 6000 Raman Shift (cm-1)

Fig. 8. Temperature dependence of the Big two-magnon response in EuBa2Cu307 crystals for temperatures higher than T N. The solid lines are fits using a model that accounts for spin-phonon interactions (after Knoll et al. 1990).

RAMAN SCATTERING STUDIES OF THE HIGH-T~ CUPRATES

529

3.1.5. Dependence on lattice parameter

Two-magnon Raman scattering studies as a function of the rare-earth ion (Tokura et al. 1990, Cooper et al. 1990) and pressure (Aronson et al. 1991, Maksimov and Tartakovskii 1994) in the cuprates have shown that J has an anomalously weak dependence on the in-plane CuO spacing, r, J ~ 1//' where n ~ 3-6, in comparison with the dependence of J on lattice parameter observed in conventional two-dimensional antiferromagnets such as KzNiF4 and K2MnF4, J ~ 1/r 12 (Jongh and Block 1975). The implications of this anomalous behavior in the cuprates can be explored by considering the relationship for the exchange interaction J in the two-band Hubbard model, which in the limit tpd << ACT is given by (Zhang and Rice 1988) J=4

AcT

+ - 2ACT 4- Up

,

(22)

where ACT is the charge-transfer gap, J is the exchange constant, tpd is the CuO hopping energy, and Ud and Up are the d- and p-band on-site Coulomb interactions, respectively. Thus, the weak dependence of J on the CuO bond length suggests either an unusually small dependence of tpd o n r (Aronson et al. 199l), or an effective cancellation of the dependencies of tpd and ACT on r (see eq. 22) (Ohta et al. 1991). 3.2. Chiral spin fluctuations

in addition to the spin-pair scattering discussed above, it has been suggested that more exotic "topological" excitations such as chiral spin fluctuations should be observable by light scattering in two-dimensional spin-½ Heisenberg systems like the cuprates (Shastry and Shraiman 1990, 1991, Khveshchenko and Wiegmann 1994). Such dynamical fluctuations of spin chirality are expected, and directly observable via Raman scattering, in spite of the fact that the ground-state expectation value of the spin chirality is zero in the insulating N~el state of the cuprates. Raman scattering from chiral spin fluctuations occurs via the scattering Hamiltonian Ha e( ~

Si " (Sj × Sk) ,

(23)

which transforms like the completely antisymmetric representation of the crystal's space group; i.e., A2g [=xy(x 2 _y2)] in the D4h-symmetry cuprates. Sulewski et al. (1991) first reported the observation of chiral spin fluctuations in the antiferromagnetic insulators, Gd2CuO4 and Pr2CuO4 (see fig. 9). In addition to exhibiting the proper A2g symmetry, the scattering observed by Sulewski et al. has a peak energy near 5J, which compares well with the chiral spin fluctuation energy estimated from spin-wave calculations. Interestingly, it has also been proposed that chiral fluctuations are responsible for the anomalous normal-state properties of the metallic cuprates (Wen et al. 1989, Y. Chen et al. 1989, Nagaosa and Lee 1990, 1991); in particular, Nagaosa and Lee (1991) predicted the

530

S.L. COOPER

6 [-----,

/

4

i ....

i

i

's

Gd~Cu04

B2 ' ~ ~

z

0

..... = l 1 0 2000 4000 6000 8000 ENERGY SHIFT ( cm .1 )

10000

Fig. 9. Broad A2g symmetry scattering in the antiferromagneticinsulator Gd2CuO4 attributed to chiral spin fluctuations (after Sulewski et al. 199l).

development of a quasielastic Raman response [Imx(~o) = Ae)F/(J + F2)], associated with closed-loop charge hopping in the N6el state, upon doping the antiferromagnetic insulating phase of the cuprates. Such a quasielastic response should also be distinguished by its appearance in the distinctive A2g (in the D4h-symmetry cuprates) scattering geometry. However, while evidence for such a quasielastic response has been observed in other doped antiferromagnetic oxides (Yoon et al. 2000), there is as yet no Raman scattering evidence for chiral spin fluctuations in the metallic phase of the cuprates. 3.3. Excitonic scattering Large-energy-shift Raman scattering in the insulating phase of the cuprates also reveals sharp electric-dipole-forbidden electronic transitions with A2g, Big, and Alg symmetries roughly 0.2eV below the 1.7eV charge-transfer gap (Liu et al. 1993a, Salamon et al. 1995, 1996) (see fig. 10). These high-energy excitations have been variously attributed to excitations involving bound electron-hole pairs stabilized by topological "hedgehog"type spin excitations (Khveshchenko and Wiegmmm 1994), intrasite d9-d 9 excitations (Liu et al. 1993a,b), and excitonic d9-dl°L excitations involving a hole transition from the Cu dx2 y2 state to a linear combination of Cu dxy and nearest-neighbor O p~ orbitals (Salamon et al. 1995, Simon et al. 1996). The latter interpretation in particular appears to

531

RAMAN SCATTERING STUDIES OF THE HIGH-T c CUPRATES

(a)

(b) I

I

I

I

I

I

¢Eu2Cu04

a

E ¢-

Sm2CuO 4 8000 10000 12000 1400016000

Raman Shift (cm-1) I

I

I

8000 10000 12000 14000 16000 Raman Shift (cm-1)

(C) A2g exciton:

Cud

Fig. 10. Large-energy-shift Raman spectra in the antiferromagnetic insulators R2CuO 4 (R=Gd, Eu, Sm): (a) A2g-symmetry exciton peak roughly 0.2 eV below the charge-transfer absorption edge (o) of several RaCuO 4 insulators. (b) Temperature dependence of the A2~ exciton in Gd2CuO4, illustrating the sharpening of the main exciton peak, and the development of two side bands (see arrows). (c) Diagram of A2g exciton involving hybridized CudxFOp~ states (after Salamon et al. 1995).

provide the best explanation for the A2g-symmetry Raman excitation, as it is consistent with the observed scaling of the A2g energy with Cu-O bond distance (Salamon et al. 1995), with estimates of 4~y-p~r exciton energies using an expanded Hubbard model involving six bands (Simon et al. 1996), and with the results of recent X-ray Raman scattering measurements of Sr2CuO2C12 (Kuiper et al. 1998). The calculations of Simon et al. (1996) further suggest that the Alg, Elg-symmetry Raman excitations, as well as the Eu-symmetry absorption peak (Falck et al. 1992, Perkins et al. 1993), are attributable to the charge-transfer d-p excitations anticipated by the standard three-band Hubbard model. At low temperatures, the A2g-symmetry excitonic peak in many insulating cuprates sharpens dramatically, revealing two sidebands that are roughly 600 c m ~ (75 meV) and 1500 c m 1 (l 88 meV) higher in energy than the main 1.5 eV peak (see fig. 10). Based upon the proximity of these energy shifts to a 600 cm 1 in-plane O mode and the exchange energy, J, respectively, these sidebands were attributed to exciton+phonon and exciton+ magnon scattering (Salamon et al. 1995). Similar sidebands on the highenergy side of two-magnon excitations in La2CuO4 were also observed in infrared

532

S.L. COOPER

absorption measurements (Perkins et al. 1993), and similarly attributed to bimagnon+ phonon emission (Lorenzana and Sawatzky 1995a,b).

3.4. Crystal-field excitations Transitions between crystal-field-split f-electron multiplet levels of various trivalent lanthanide ions (R) have also been observed by Raman scattering in the insulating phase of high-T~ cuprates (Heyen et al. 1991, Ruf et al. 1992, Dufour et al. 1995, Sanjurjo et al. 1995, Jandl et al. 1995, Cardona 1999), particularly those containing Nd 3÷ (419/2), Pr 3+ (3H4) , and Sm 3+ (6Hs/a) ions. These studies have revealed both 'pure' crystal-field excitations in R2CuO 4 systems, as well as crystal-field excitations in RBa2Cu306 systems that derive their Raman intensities from a strong magneto-elastic coupling with phonons. For a more detailed discussion of these studies, the reader is referred to the recent review article by Cardona (1999).

4. Raman scattering in the "normal" metallic phase

4.1. ab-plane electronic scattering in the optimally doped and overdoped regimes' One of the most remarkable features of the Raman scattering spectra of high-Tc cuprates in the "normal" metallic phase is the presence of a broad, flat scattering "continuum" of electronic Raman scattering that extends from zero energy to energies as large as 1 eV (see fig. 1 l). This normal-state continuum was probably first observed in YBa2Cu307 films (Bozovic et al. 1987) and polycrystals (Lyons et al. 1987, Bazhenov et al. 1987), then later in single crystals of YBa2Cu307 (Cooper et al. 1988a, Thomsen et al. 1988, Sugai et al. 1988, X.K. Chen et al. 1993), La2 xSrxCuO4 (Katsufuji et al. 1993, X.K. Chen et al. 1994a), Bi2Sr2CaCu208 (Slakey et al. 1990a, Staufer et al. 1992, Kendziora and Rosenberg 1995), T12Ba2Ca2Cu3Ot0 (Krantz et al. 1989, Hoffmann et al. 1994), and T12Ba2CuO6 (Kang et al. 1996, Gasparov et al. 1997). Single-crystal studies of YBa2Cu3Ov also revealed a strong coupling between the electronic Raman continuum and certain optical phonons, manifested as strong Fano phonon lineshapes (Cooper et al. 1988a, Thomsen et al. 1988). While the Raman continuum has generally been assumed to be of import to the unconventional normal-state properties of the cuprates (Varma et al. 1989, Cooper and Klein 1990, Virosztek and Ruvalds 1991, 1992), the origin of this scattering is still uncertain, and has been attributed variously to strong electronelectron scattering in the presence of Fermi surface nesting (Virosztek and Ruvalds 1991, 1992), strongly correlated, incoherent electronic scattering associated with a MottHubbard system (Shastry and Shraiman 1990), electronic scattering from spin fluctuations (Branch and Carbotte 1995), the effects of inelastic and interband electronic scattering (Rashkeev and Wendin 1993), mass fluctuations between different parts of the cuprates' complex Fermi surface (Krantz and Cardona 1995), and chiral fluctuations in the normal state (Nagaosa and Lee 1990, 1991).

RAMAN SCATTERING STUDIES OF THE HIGH-T~ CUPRATES

533

i BaxKI.xBiO

3

YBazCu307-y i

iR \ Ba0.5oKo.zsB=Oy i~gao.47Ko.39BiOv

~2

J

wI

Zl

(Lal-xSrx)zCuO4

I

BizSrzCal-xYxCuzOs.y

li l

i J I!X=O

u t~ ,

ii

kf,s

1000

,A . . . .

X -0

......

• ..,.~

ll~, .i

l

0

/

X=0.058

I

I

I

I

I

I

2OOO

I

I

I

l

-

.

I

300O

0

I000

200O

3OOO

t,O00

ENERGY SHIFT (cm-I) I

l

i

l

i

i

I

i

i

Bi 2 Sr 2 Ca Cu 20 s R. HACKL ,..DATA (XX) ¢0

CO

z

NFL a=oA

i I

0

I

2000

I

]

4000

I

X

6000

J

I

8000

i,

10000

CO(cm "1) Fig. 11. (top) Broad, normal-state Raman continua in various oxide materials at room temperature (after Sugai et al. 1989). (bottom) Room-temperature Raman continuum in Bi2Sr2CaCu208 fit to a response involving the malginal and nested Fermi liquid scattering rate form (see eq. 26) (after Virosztek and Ruvalds 1992),

534

S.L. COOPER

In the absence of a specific microscopic picture, the observed normal-state Raman continua of the cuprates have generally been modeled with a "collision-dominated" scattering response (see sect. 2.5.3), and associated with overdamped fluctuations in the strongly correlated normal-state fluid (Klein et al. 1989, Cooper and Klein 1990, Slakey et al. 1991), d2o" df2dco

AcoF [n(co)+ 1] ~ 2 + f f 2 ,

(24)

where A is a scattering amplitude and F = 1/T is the scattering rate. To adequately describe the flat scattering response of the "continuum" in the cuprates, the response function in eq. (24) suggests a scattering rate of the form (Klein et al. 1989, Cooper and Klein 1990, Virosztek and Ruvalds 1991, 1992) F((D, T ) = aV'/(.o 2 q- ~ 2 T 2

[h = k B = 1].

(25)

Note that eq. (25) is simply the phenomenological normal-state electronic scattering rate proposed by Varma et al. (1989) for "marginal" Fermi liquids (MFL), which has also been shown to arise from electron-electron scattering on a nested Fermi surface (Virosztek and Ruvalds 1991, 1992) and spin-fluctuation scattering (Moriya et al. 1990, Sokol and Pines 1993). As shown in the bottom panel of fig. 11, the phenomenological expression in eq. (24), combined with the MFL scattering rate in eq. (25), fits the normal-state continuum in Bi2SrzCaCu208 over a wide frequency range. Furthermore, the Alg-symmetry Raman continuum has been found to exhibit a MFL-like temperature dependence (Slakey et al. 1991, Reznik et al. 1993), suggesting that the "continuum" in this symmetry is associated with scattering from conduction electrons. On the other hand, the temperature dependence of the Big-symmetry continuum deviates from the MFL form (Reznik et al. 1993), possibly indicating that processes not associated with the conduction electrons contribute to the scattering response in this symmetry (see sect. 5.3.1 and X.K. Chen et al. 1993). Additional evidence that the Raman continuum in the cuprates is associated with scattering from the conduction electrons, at least in the Alg and B2g symmetries, is suggested by the reasonably close connection between scattering rates extracted from the measured continuum response using eqs. (24) and (25), and transport scattering rates. For example, fits to the Raman scattering data of YBa2Cu306+x and Bi2Sr2CaCu208 using eq. (24) and a scattering rate form only slightly modified from eq. (25) (Hackl et al. 1996, Opel et al. 2000) show that the Bzg-continuum scattering rate compares well to the scattering rate estimated from transport measurements, both as a function of temperature and doping. By contrast, the scattering rates extracted from fits to the Big-symmetry continuum are in substantial excess of the transport scattering rate deduced from infrared (Tanner and Timusk 1992, Orenstein et al. 1990) and transport measurements. This suggests that transport lifetimes, at least in Bi2Sr2CaCu2Os, are dominated by scattering processes near the (kx,ky)=(ijr/a,-t-Jr~a) [i.e., B2g] parts of the Fermi surface, while

RAMAN SCATTERING STUDIES OF THE HIGH-T0 CUPRATES

535

scattering processes near the (kx,ky)-{(0,-4-~/a), (~:r/a, 0)} points [i.e., Blg] are shortlived and do not significantly influence transport lifetimes. Interestingly, however, Raman measurements of the Big scattering rate in overdoped La2 xSr~CuO4 by Naeini et al. (1998) not only imply that this contribution to the scattering rate is important to the transport properties, but also indicate that the temperature dependence of the Big scattering rate is more consistent with a nearly antiferromagnetic Fermi liquid (NAFL) model (Pines 1997) than the MFL picture. This is consistent with other measurements indicating a cross-over to more conventional Fermi-liquid behavior in the overdoped regime (see fig. 1) (Batlogg et al. 1994). A fundamental question regarding the normal-state continuum in the cuprates concerns the nature of the Raman scattering vertex (i.e., y in eqs. 15 and 19) associated with this process. Krantz et al. (1995) and Krantz and Cardona (1995) provided perhaps the best evidence that the normal-state continuum in the cuprates is attributable to mass fluctuations between different parts of the complex cuprate Fermi surface (see discussion in sect. 2.5.2). They calculated the electronic Raman scattering efficiency for YBa2Cu307 in the "collision-dominated" regime, using a modification of the squared Raman scattering vertex in eq. (19) given by

1.2 IyI 2 - ~ N ( O ) ( I E I • (~ 1 - / , - 1 ) ) "

ES]2),

(26)

where N(0) is the one-electron density of states at the Fermi level, and where oneelectron band structure calculations (Andersen et al. 1991, 1994) were used as the basis for determining the effective mass tensor, g. Significantly, these calculations appear to reproduce the relative intensities of the electronic continuum in different scattering geometries (see fig. 12), although the calculated absolute magnitudes are roughly an order of magnitude too large (Krantz et al. 1995, Krantz and Cardona 1995). Perhaps the most important question related to the broad Raman continuum in the cuprates, however, concerns its relevance to the unusual normal- and superconductingstate properties of the cuprates. It is worth noting in this regard that broad, flat electronic scattering continua are not unique to the cuprates, and indeed have been observed in such diverse materials as BaxK~ xBiO3 (Sugai et al. 1989) (see fig. 11), Sr, Lal-xTiO3 (Katsufuji and Tokura 1994), EuB6 (Nyhus et al. 1997), Al_xA~xMnO3 (Yoon et al. 1998), and La0.sSr0.sCoO3 and Ca0.sSr0.sRuO3 (Bozovic et al. 1994). The ubiquity of electronic Raman continua in strongly correlated materials supports the argument that such scattering should arise rather generally in materials with both a complex Fermi surface and strong electronic scattering (see discussion in sect. 2.5.2). Moreover, the prevalence of electronic Raman continua in strongly correlated materials raises obvious questions concerning the significance of this response to the distinctive normal-state properties of the high-Tc superconductors. However, Leggett (1998) recently suggested that it is not the mere presence of a strong mid-infrared dielectric response, but rather the unique interplay between this response and the low dimensionality of the cuprates that contributes to high-Tc superconductivity. Therefore, the importance of the broad normal-

536

S.L. COOPER 50

E O ""

40

"7,

E 30

L~

2v4x,2V+J

..--.

20

tO

•-~

o

10

o - -

UJ

0 200

400

600

800

Raman Shift (cm "1)

Fig. 12. 300K Raman spectra of a YBa2Cu?O 7 single crystal in different polarizations. The spectra have consecutive offsets of (0, 1,2,..., 8) x 2.5 (after Krantz et al. 1995).

state Raman continua in the cuprates to high-To superconductivity remains unresolved (see also sect. 5.3.5). 4.2. c-axis electronic scattering In addition to their remarkable ab-plane properties, the high-To cuprates are among the most anisotropic material systems known, exhibiting resistivity ratios in the range p~./p~, ~ 102-105 (Cooper and Gray 1994). By comparison, the layered dichalcogenide superconductors 2H-NbSe2, 4Hb-TaS2, and 2H-NbS2 have resistivity anisotropies of pc/pab~ 101-103. The large anisotropy of the layered cuprates manifests itself in a number of normal- and superconducting-state properties, including the temperaturedependent resistivity (Ito et al. 1991), which is metallic in the ab-plane but often exhibits a "semiconducting-like" temperature dependence in the c-direction, and the optical conductivity (Uchida et al. 1991, Cooper et al. 1993a,b). These properties provide strong evidence that c-axis transport in the under- and optimally doped high-To cuprates is incoherent. c-axis polarized [(EbgS) = (z,z)] electronic scattering has been reported in YBa2Cu306 ~x (Reznik et al., 1992, Cooper et al. 1993a,b, Nyhus et al. 1994), T1Ba2CaCu207 (Mihailovic et al. 1993), T12Ba2CaCu208 (Mihailovic et al. 1993), and Bi2Sr2CaCu208 (Liu et al. 1999). In YBa2Cu306+x, for example, c-axis electronic scattering is characterized by a collision-dominated scattering response, Imx(co)= A~oF/(co2+ F2), with a constant scattering rate F = 1/r ~ 600 cm 1, and an intensity that increases roughly

RAMAN SCATTERING STUDIES OF THE HIGH-Tc CUPRATES I

I

x-(0.85

I

I

0.75 1 0.5 1 0.37

YBa Cu 0

1 ~ ~t

0.25

+

"\.

¢°o /

2 3 6+x (%%) = (Z,Z)

•

/

-,-++.._+ T+5K

"',ooo

+ 1300

400

537

2200

ooo ,ooo

3100

4000

Energy Shift (cm -1) 3

'i 0.$

2

4

6

B

I0

Fig. t3. (top) Room-temperature, c-axis polarized [(el, es) = (z,z)] electronic scattering in YBa2Cu30~,,., as a fimction of doping (x); (inset) The T = 5 K c-axis Raman spectrum of YBa2Cu306.93 , with a fit to a collisiondominated response function that has been shifted by the 500 cm ~ apical 0(4) phonon energy (after Nyhus et al. 1994). (bottom) Calculated Raman spectrum of a single Einstein phonon coupled to a background continuum [dashed lines] (after Itai 1992).

linearly with carrier density (Cooper et al. 1993a,b, Nyhus et al. 1994) (see fig. 13). This incoherent c-axis scattering response reflects the fact that the carriers scatter many times in the plane before hopping between planes; i.e., 1~tab >> 1/Tc. There is some evidence that phonons may participate in the c-axis continuum scattering response of YBa2Cu306+x, via an "activated hopping" process (Nyhus et al. 1994). For example, the c-axis electronic scattering response in YBa2Cu306+x "turns on" for energies greater than the 500cm i apical 0(4) phonon frequency (see fig. 13), suggesting that this phonon is necessary for momentum conservation in the interlayer scattering process. Indeed, the process of Raman scattering from electron-hole pairs

538

S.L. COOPER

with the emission of a phonon (bottom, fig. 13) has in fact been calculated by Itai (1992), and the results of these calculations describe the observed c-axis continuum response in YBa2Cu306+~ well. Consequently, these results suggest the interesting possibility is that the strong c-axis electronic continuum in YBa2Cu306+x (and perhaps the Tl-based cuprates) betrays a transport mechanism in which c-axis phonons assist the hopping of carriers between the CuO layers. This would likely result from a dynamic modulation of the O(4)pz-Cu(1,2)d3z2 i overlap integral by the 500cm 1 apical 0(4) optical phonon, which consequently modulates the interbilayer hopping rate in YBa2Cu306+x. The sensitivity of interlayer coupling in both La2 xSrxCuO4 and YBa2Cu306 ~x to the hybridization between the apical 0(4) Pz orbital and the in-plane Cu states is supported by X-ray absorption measurements of C.T. Chen et al. (1992) and resonance Raman scattering experiments of Heyen et al. (1990, 1992), and is a direct consequence of the weak interlayer hopping rates in the cuprates. Importantly, phonon-assisted c-axis hopping has been proposed as a possible contribution to the "semiconductor-like" c-axis resistivities observed in many of the cuprates (Kurnar and Jayannavar 1992, Nyhus et al. 1994, Rojo and Levin 1993, Zha et al. 1996). However, a more recent Raman study and analysis of the c-axis electronic continuum in YBa2Cu308 suggests that the c-axis Raman continua in Y-based cuprates are not strongly influenced by the presence of the CuO chains (and hence may not be directly related to c-axis transport) or by phonon scattering (Quilty et al. 2001). The exact origin of the c-axis continua in the cuprates, therefore, remains uncertain. Interestingly, Wu and Carbotte (1997) have calculated the c-axis electronic Raman scattering response for a plane-chain bilayer coupled via single-particle tunneling, including both intraband and interband contributions. These results predict that interband scattering provides the dominant c-axis Raman scattering response in the limit of small plane-chain coupling, although intraband and interband scattering contributions are not well separated due to Coulomb screening. These calculations also suggest that "pseudogap" development, which in this model is associated with the splitting between plane and chain bands, should be observable in the c-axis Raman scattering spectrum as well as the c-axis optical conductivity (Homes et al. 1993). Interestingly, Quilty et al. (2001) recently observed a subtle depletion of spectral weight with decreasing temperature in the normal-state c-axis Raman continua of both YBa2Cu308 and underdoped YBa2Cu3 06 +x (Tc ~ 80 K), which they attributed to pseudogap development. However, as the origin of the c-axis Raman continuum remains uncertain, the mechanism by which it is affected by pseudogap formation is not clear. 4.3. Raman scattering in the underdoped cuprates The underdoped regime of the high-Tc cuprates is currently perhaps the most poorly understood and interesting phase regime. This phase is thought to be characterized by at least two cross-over temperatures (see fig. 1) (Batlogg and Emery 1996, Emery, Kivelson and Zachar 1997): (1) a temperature, TT, below which charge inhomogeneities (stripes) and short-range antiferromagnetic correlations may be present in the CuO2 planes; and

RAMAN SCATTERING STUDIES OF THE HIGH-To CUPRATES

539

(2) a cross-over temperature, T~, below which "pseudogap" behavior is observed (RossatMignod et al. 1991, Warren et al. 1989, Walstedt et al. 1990, Loeser et al. 1996, Ding et al. 1996, Marshall et al. 1996) and pairing correlations may occur above To. Raman scattering studies of the underdoped phase of the cuprates have focused primarily on several related issues: (a) the possible persistence of antiferromagnetic correlations into the metallic phase; (b) evidence for pseudogap development in the Raman scattering spectra, and the relationship of this pseudogap to the superconducting and antiferromagnetic phases; and (c) evidence for normal-state pairing of quasiparticles above To. 4.3.1. Presence of antiJerromagnetic correlations As discussed in greater detail in sect. 3.1.3, there is some evidence (Lyons et al. 1991, Blumberg et al. 1994, Rfibhausen et al. 1997, 1999) for the persistence of twomagnon Raman scattering in metallic YBa2Cu306+x (for 0.3 <x <0.7), YBazCugQ, Yl xPrxBa2Cu307, and BizSr2CaCu2Os, suggesting that antiferromagnetic fluctuations with a spatial extent of several lattice constants are present well into the underdoped phase of the cuprates. Moreover, Rfibhausen et al. (1997, 1998) have reported that overdamped two-magnon excitations in the metallic phase of underdoped and optimally doped Yt-xPrxBa2Cu307 and Bi2SrzCaCu2Q are influenced by superconductivity, suggesting a coupling between the itinerant electronic states in the metallic phase and the remnant short-wavelength spin degrees of freedom. Note, however, that two-magnon scattering has been observed to disappear upon further doping into the overdoped phase of Bi2SrzCaCu208 (Rfibhausen et al. 1999), suggesting that antiferromagnetic correlations are less than twice the lattice parameter in this phase regime; the authors suggest that this behavior reflects a change in character from a doped antiferromagnetic phase (Dagotto 1994, Plakida et al. 1997) in the underdoped regime of the cuprates, to an antiferromagnetic Fermi liquid phase (Scalapino 1995, Anderson 1997, Pines 1997) in the overdoped regime. A potentially interesting area of investigation for Raman scattering in the underdoped cuprates concerns the study of ordered charge and spin structures; i.e., "stripes", observed in neutron scattering and electron diffraction measurements of the cuprates (Tranquada et al. 1996) and nickelates (C.H. Chen et al. 1993, Tranquada et al. 1994). While Raman scattering from the charge and spin dynamics of such ordered structures has not yet been reported in the cuprates, Yamamoto et al. (1998) and Blumberg et al. (1998a) recently reported Raman scattering evidence for double-spin excitations associated with striped spin domains below the charge-ordering temperature of Lal.67Sr0.33NiO4. These results offer encouragement that studies of the inhomogeneous phases in underdoped cuprates might also be fruitful. 4.3.2. Evidence for a pseudogap The first evidence for a "pseudogap" in the Raman scattering response of the cuprates was actually inferred from the anomalous temperature dependences of Raman- and infraredactive phonons (Litvinchuk et al. 1992). More recently, evidence for a pseudogap has

540

S.L. COOPER

T= 300 K 150 K 80 K 4O K

To= 88 K (underdoped)

~

I ~''g I // ~M,.,~. /

/,_/"~ ~

585 cm -1 peak

-" 0

pseudogap 200

400

600

Raman Shift (cm4)

800

Fig. 14. The B~ symmetry electronic Raman scattering response of underdoped Bi2Sr2CaCu208 as a function o f temperature, illustrating evidence for pseudogap formation below ~450 cm 1, and the development of a peak at 585 cm -~ that is associated with normal-state fermion pair formation (after Blumberg et al. 1998b).

been inferred from measurements of electronic Raman scattering in underdoped cuprates (X.K. Chen et al. 1997, 1998b, Nemetschek et al. 1997, Blumberg et al. 1997b, Quilty et al. 1998, Naeini et al. 1999, Hewitt et al. 1999, Opel et al. 2000), but the results and interpretations of different groups are not yet in agreement on all issues. On one hand, there appears to be general agreement that underdoping in YBa2Cu306 +x (X.K. Chen et al. 1998b), La2~xSrxCuO4 (Naeini et al. 1999), and BizSr2CaCu208 (Hewitt et al. 1999) leads to a reduction of electronic scattering strength in the Big scattering geometry. As the Big-symmetry Raman vertex function preferentially samples Fermi surface regions (kx,ky) near {(O,-4-~/a),(+Jr/a,O)} (see eq. 17 and fig. 2), this reduction of B i g electronic scattering intensity is consistent both with the development of a pseudogap near (kx,ky) = {(0,-4-3r/a), (+Jr~a,0)}, and with angleresolved photoemission spectroscopy (ARPES), which find that the pseudogap has a dx2_y2 (B~g) momentum dependence (Ding et al. 1996, Loeser et al. 1996). There is less uniformity among different groups regarding the temperature dependence of the electronic Raman scattering intensity in the "pseudogap" phase. Nemetschek et al. (1997) inferred the presence of a pseudogap from the temperature-dependent reduction of Bzg-symmetry electronic scattering intensity in underdoped BizSr2(Ca0.62Y0.38)Cu208 and YBazCu306+x. Significantly, however, this group did not observe a similar suppression of electronic scattering in the Big spectrum. This result appears to be at odds with the dx2_y~symmetry pseudogap inferred from ARPES measurements (Ding et al. 1996, Loeser et al. 1996), since the Bzg-symmetry Raman vertex function preferentially samples Fermi surface regions (kx,ky) near (dcJv/a,+zUa) (see eq. 17 and fig. 2) - by contrast, the large gap regions of a dx2_y~-symmetry pseudogap should manifest themselves primarily in the Big-symmetry electronic Raman scattering response. X.K. Chen et al. (1997) also saw no evidence for a temperature-dependent reduction in Big-symmetry scattering that would be expected from the temperature development of a dx~ y~ pseudogap. On the other hand, Blumberg et al. (1997b, 1998b), Qnilty et al. (1998), and Hewitt et al. (1999) observed a temperature-dependent suppression of electronic scattering in the Big-

RAMAN SCATTERING STUDIES OF THE HIGH-Tc CUPRATES

541

symmetry electronic spectrum of Bi2Sr2CaCu208 that is consistent with the development of a d,2 _y2 symmetry pseudogap (see fig. 14) below T* > Tc. The disparity among various Raman results in the underdoped phase is puzzling and requires further investigation.

4.3.3. Evidence for normal-state pairing Among the principal explanations for the development of a normal-state "pseudogap" in the underdoped cuprates are (i) the pairing of spinons above Tc (Wen and Lee 1996, Anderson 1987), and (ii) the formation of normal-state fermion pairs (Emery and Kivetson 1995, Ranninger and Robin 1996). The former interpretation appears to be favored by several Raman scattering studies (X.K. Chen et al. 1997, Naeini et al. 1999, Hewitt et al. 1999, Opel et al. 2000) which find the magnitude of the pseudogap to be given by Eg ~ J ~ 100 meV, where J is the antiferromagnetic superexchange coupling constant, but which find no evidence for a connection between the pseudogap and superconducting gap. On the other hand, the latter interpretation does appear to be supported by ARPES measurements (Ding et al. 1996, Loeser et al. 1996), which indicate that the pseudogap (a) has dx2y2 symmetry, (b) has a magnitude on the order of the superconducting gap, Eg ~ A, and (c) evolves continuously into the superconducting state. Significantly, there are several Raman scattering studies of the underdoped cuprates that have also been cited as evidence for some form of normal-state pairing. For example, Slakey et al. (1990b) observed that a 500 cm 1 Big-symmetry peak normally associated with pair breaking in the optimally doped superconducting phase (see sect. 5.1) persists well above To in underdoped YBa2Cu306+x. However, two recent measurements of underdoped YBa2Cu306+x found no evidence for a Big-symmetry Raman peak above T~ (Nemetschek et al. 1997, X.K. Chen et al. 1997). Several groups (Blumberg et al. 1997b, Quilty et al. 1998, Hewitt et al. 1999) have reported that the suppression of low-frequency electronic scattering with decreasing temperature in underdoped Bi2SrzCaCu2Os, ostensibly due to pseudogap formation, is accompanied by the development of a 585 cm -1 (73 meV), Big (dx2 y2) symmetry peak above Tc (see fig. 14). Blumberg et al. (1997b) attributed the development of this peak to precursor pairing of doped holes into Big-symmetry states above Tc. This interpretation is consistent with the model of Emery, Kivelson and Zachar (1997), which predicts that bound fermion pairs in the pseudogap phase of the cuprates should have a Raman-active, charge-0, spin-0 collective excitation with Big symmetry and an energy, ~ A ~ ~ 70 meV, based upon neutron scattering measurements of the spin gap, As (= 40 meV, Fong et al. 1995, 1996). However, more recent Raman measurements question this interpretation of the 585 cm 1 peak: measurements of the ~585 cm 1 Ble.symmetry peak in oxygendeficient Bi2SraCaCu2Os by Hewitt et al. (1999) demonstrated that it softens by 3.8% upon isotopic substitution of 160 by 180. Furthermore, these researchers found that this peak is not observed in samples underdoped by Y-substitution, rather than by oxygen-deficiency. Based upon these observations, Hewitt et al. (1999) concluded that

542

S.L. COOPER

the 585 c m I Big-symmetry peak in BizSrzCaCu208 is vibrational in origin, probably arising from phonons activated by oxygen defects.

5. Electronic scattering in the superconducting state

5.1. Overview It was first pointed out by Abrikosov and Fal'kovskii (1961) that the superconducting gap should manifest itself in the electronic Raman scattering response as an abrupt onset of scattering intensity at the pair-breaking energy, ho)=2A. Extensions of this initial calculation of the Raman scattering cross section of a superconductor later included the effects of Coulomb interactions (Abrikosov and Genkin 1974, Cuden 1976), and considered the influence of Fermi surface anisotropy (Abrikosov and Genkin 1974). Most early theoretical treatments of Raman scattering in superconductors assumed that the optical skin depth 6 was less than the coherence length ~0 of the superconductor. However, the first experimental observations of Raman scattering from superconducting gap excitations were actually made in conventional s-wave superconductors for which ~0 << 6, including the layered dichalcogenide compound 2H-NbSe2 (Sooryakumar and Klein 1980, 1981), and the A15 compounds Nb3Sn and V3Si (Dierker et al. 1983, Hackl, Kaiser and Schicktanz 1983, Hackl and Kaiser 1988, Hackl, Kaiser and G1/iser 1989) (see fig. 15). In many of these studies, it was reported that Raman intensity at 2A in certain symmetries was gained at the expense of intensity in a Raman-active phonon (Sooryakumar and Klein 1980, 1981, Hackl and Kaiser 1988, Hackl, Kaiser and G1/iser 1989), suggesting that the "gap modes" in these symmetries derive their Raman activities from a coupling between phonons and electronic excitations (Balseiro and Falicov 1980, Littlewood and Varma 1981, 1982, Browne and Levin 1983, Lei et al. 1985, Zeyher and Zwicknagel 1990). In Nb3 Sn and V3 Si, however, an onset of electronic scattering at 2A is also observed in scattering geometries for which there are no Raman-active phonons, and therefore can only arise from direct Raman scattering from quasiparticle pairs. A BCS-based calculation of Raman scattering from quasiparticle pairs for the case ~0 << 6 (qvv << 2A) was first presented by Klein and Dierker (1984) for the limit q -+ 0; this calculation incorporated the effects of both Coulomb screening and the attractive pairing interaction, and examined the polarization dependence arising from gap anisotropy. Abrikosov and Fal'kovskii (1988) later extended this calculation to include the first correction to the scattering cross section that is nonzero in the parameter qvF/A. More recent calculations of the electronic Raman scattering cross-section in the superconducting state examined the effects of a k-dependent energy gap with line nodes (Fal'kovskii and Klama 1990), impurity scattering (Devereaux 1992, 1993), strong electron-phonon coupling in the presence of anisotropy (Kostur 1992), and final-state interactions governing the formation of excitonic and electron-pair bound states (Monien and Zawadowski 1990, Zawadowski 1990).

RAMAN SCATTERING STUDIES OF THE HIGH-To CUPRATES

2

f

Egv~--------~ !

.o

543

t-Nb3Sn I ,

_

,o

19K

2 !

~ 2

• I

. ~

13K -

t~ ow

C:

.c I

"E 2

•" " •

6K °

i¢1

~

0

100

200 frequency ~

~

0 100 (cm-1 )

L

~

200

300

Fig. 15. Raman scattering from superconducting gap excitations in Nb3Sn as a ftmction of symmetiy and temperature, showing the abrupt onset of scattering for h ~ ) 2 A (~ 50cm 1) below T~, and the presence of gap modes (after Hackl et al. 1989).

In the high-Tc cuprates, Raman scattering from quasiparticle pair excitations was first observed in polycrystalline samples as a broad scattering peak near 400 c m I (Lyons et al. 1987, Bazhenov et al. 1987). The development of this "pair-breaking" peak in the electTonic Raman scattering spectrum below Tc was later corroborated in measurements of single-crystal YBa2Cu307 (Cooper et al. 1988a, Thomsen et al. 1988), Bi2Sr2CaCu208 (Kirillov et al. 1988, Yamanaka et al. 1988, Slakey et al. 1990a), T12Ba2Ca2Cu3010 (Krantz et al. 1989, Hoffmann et al. 1994), La2_xSrxCuO4 (X.K. Chen et al. 1994a),

544

S.L. COOPER

YBa2Cu408 (Donovan et al. 1995), T12Ba2CuO6 (Nemetschek et al. 1993, Kang et al. 1996, Gasparov et al. 1997), and HgBa2CaCu206 (Sacuto et al. 1997, Zhou et al. 1997) (see fig. 16). Significantly, several key features of the superconducting-state Raman response of high-Tc cuprates betray an unconventional gap in these materials. First, while there is a substantial suppression of low-frequency electronic scattering in the superconducting state of the high-To cuprates, there is no well-defined gap in the scattering spectrum down to the lowest measured temperature and energy (Cooper et al. 1988a), a feature which distinguishes the high-To cuprates from the complete suppression of scattering observed below 2A in conventional superconductors such as Nb3Sn (Dierker et al. 1983) and V3Si (Hackl, Kaiser and Schicktanz 1983). Second, the putative "pair-breaking" or "coherence" peak observed in the electronic Raman scattering response of optimally doped high-Tc superconductors has several distinctive properties: (i) it has a substantially larger energy, -8kBTc, than that expected in a BCS superconductor, 2ABcs ~ 3.5kB Tc; (ii) it is essentially temperature independent below To, a feature also not expected in a BCS superconductor; and (iii) it has a roughly 30% higher value in the Blg (dx2 y2) scattering geometry than in the Alg and B2g (dxy) scattering geometries in most cuprates, e.g., YBa2Cu307 (Hackl et al. 1988, Cooper et al. 1988b), Bi2Sr2CaCu208 (Staufer et al. 1992, Devereaux et al. 1994a, Kendziora and Rosenberg 1995), La2 xSrxCuO4 (X.K. Chen et al. 1994a), and T12Ba2CuO6 (Kang et al. 1996, Gasparov et al. 1997). Finally, the superconducting state Raman responses of most cuprates, including Bi2Sr2CaCu208 (Devereaux et al. 1994a), La2 xSrxCuO4 (X.K. Chen et al. 1994a), TlzBa2CuO6 (Gasparov et al. t997), and HgBa2Ca2Cu308 (Sacuto et al. 2000), exhibit a o)3 frequency dependence in the Big (dx2__y2) scattering geometry, but a linear-in-co dependence in the Alg and B2g scattering geometries. Devereaux and coworkers (Devereaux et al. 1994a,b, Devereaux and Einzel 1996a) offered the first theoretical evidence that many of these unconventional properties could be identified as the electronic Raman scattering response of a superconductor with a dx2 y2 pairing state. To understand the basis for this interpretation, it is first useful to consider in more detail the general theory of electronic Raman scattering in a superconductor. 5.2. Theory of Raman scattering in superconductors 5.2.1. The Raman scattering cross-section The electronic Raman scattering cross-section in the superconducting state of a superconductor is derived by considering the relationship (Klein and Dierker 1984) d2o

an

_

cos ns ~

nZ .r

<[2>>6(Ef-Ei-hco),

(27)

r 2 cos A n

wi ni [n(co) + l] Im)f~(q, w),

where the Raman response function X~(q,w) is evaluated using the Bardeen-CooperSehrieffer (BCS) ground state. Klein and Dierker (1984) performed this calculation

RAMAN SCATTERING STUDIES OF THE HIGH-T~ CUPRATES

545

~o

., i";."

. i!.,'

i'i i"!~ .'.. ~ :., ::;':,.

Alg

i

~d

B

"

:;,.,

i':

lg -.

"..',

"

•

'i

~ i ~ O2g -"x~2':'~q "?~ ,~,.i .. -; +:) .... ,

0

250

50O

co [ c n i I ] 1.1

Tc=BOK/ TI-2201 ~

' ' ~'~

1,2 1,0 |

i/[

(430 cm-t)

1.0

o.a ;~0

400

0,11

600

linear in ~ f~ Wave numbers (c~,-n-t)

B2g E Big

o

,

2

~o

~

2

Raman Shift

30

,

2

3

(Amax)

Fig. 16. Symmetry dependence of electTonic Raman scattering from superconducting gap excitations in various high-Tc cuprates, including optimally doped (top) Bi2Sr2CaCu208 (after Devereaux and Einzel 1996a), (middle) single-layer T12BazCuO6 (after Gasparov et al. 1997), and (bottom) La2 xSr~CuO4 (X.K. Chen et al. 1994a). Note the 0)3 low-frequency dependence of the Big scattering response in these materials, but the linear-in-e) dependence in Atg and Big scattering responses.

546

S.L. COOPER

(a)

ZAv =

Au ~

7k

Fig. 17. (a) Polarization bubble diagrams for computing the gauge-invariant Raman scattering response of superconductors, x~(q, co) = XAy(q,co),where A and y are the renormalized and bare Raman scattering vertices, and the two solid lines represent the excitation of quasiparficle pairs. (b) Self-energy corrections to the Raman scattering vertex, including corrections due to pairing correlations (middle diagram on right-hand side) and Coulomb screening within the RPA approximation (final diagram on right-hand side). including the effects of pair breaking and Coulomb screening within the Random Phase Approximation (RPA) (Pines 1963): the Raman scattering response function is associated with a susceptibility, X~(q, 6o) = ZaZ(q, 6O), consisting of renormalized and bare Raman scattering vertices, A and y, respectively, and a pair of Green's function lines associated with the excitation of quasiparticle pairs (see fig. 17). The resulting expression for the Raman scattering response function is XIsp(q,(9) = XAY(q, 09) = xw(q, Oo)

X~l(q, ( - - co) + X21(q,o)) - -

Xl~(q, co)

m l(q, co)

l

~

)

'

(28)

where the first term on the right-hand side is the "bare" Raman scattering response, the second term reflects transverse screening, and the final term is associated with longitudinal screening, where e(q, oo) = 1 - Vqxll(q, 6O) is the dielectric function and Vq = 4JreZ/q 2 is the Coulomb interaction matrix element. The longitudinal screening term is of order O(qQe) (Devereaux and Einzel 1995), and therefore can be ignored when considering the q ~ 0 Raman scattering process. Consequently, under the conditions that (a) the superconducting gap A is much smaller than the Fermi energy, A << EF, and (b) the momentum transfer of light q is small compared to the Fermi momentum, the imaginary part of the (T = 0) electronic Raman scattering response function for a superconductor, including Coulomb screening effects, can be written (Klein and Dierker 1984, Devereaux et al. 1994b, Devereaux and Einzel 1995) as I m x ~ ( q , 6o) = Im Xzy

Xll J = Im

IYnkl2 £k>

(Y'k'~/'}2<,~k> '

where (. - .) represents an average over the Fermi surface, Ynl, is the bare Raman scattering vertex for band n in the non-resonant limit given in eq. (15), and ,~k is the imaginary part of the Tsuneto function for T = 0 (Tsuneto 1960): 4 ]z~k 12

;~k =

co ~

[h = 1].

(30)

- 4 IAk 12

Note, from eq. (30), that in an s-wave superconductor, the unscreened superconductingstate Raman scattering response, ocImxyT, is zero for co<2A, but has a scattering

RAMAN SCATTERING STUDIES OF THE HIGH-T~ CUPRATES

547

peak at CO=2A due to quasiparticle scattering above the energy gap. In the infrared conductivity, by contrast, spectral weight suppressed below 2A in the superconducting state appears as a delta function at co = 0, reflecting absorption by the superfluid (Tanner and Timusk 1992). The different frequency dependences of the Raman and infrared responses in the superconducting state derive from the different BCS coherence factors (Tinkham 1975) associated with the respective Raman scattering (Type-I) and infrared (Type-II) Hamiltonians, HR - A • ~ 1 . A and HIR ~ A .p. 5.2.2. Coulomb screening effects Coulomb screening has several important consequences for the Raman scattering response in superconductors. First, it minimizes the effects of pairing correlations (middle diagram in fig. 17b) on the low-frequency Raman scattering spectrum by pushing the zerofrequency Goldstone mode, associated with phase fluctuations of the order parameter, up to the plasma frequency COp(Klein and Dierker 1984, Devereaux and Einzel 1995, Strohm and Cardona 1997a). Second, screening has a strong influence on the scattering response in certain symmetries. Devereaux et al. (1994a,b) [see also Kostur (1992)] pointed out that in tetragonal (D4h) crystals, the second, Coulomb screening, term in the brackets of eq. (29) vanishes due to symmetry in L = 2 scattering symmetries such as Big and B2g. This is because the Tsuneto function ,~k (oc A2) has the full symmetry of the space group, Alg, and therefore the factor (gnk;.k) in the second term of eq. (29) vanishes in a tetragonal superconductor unless the Raman vertex Ynk also has Alg symmetry. Thus, the gauge-invariant, T = 0 Raman scattering response in the Big and B2g scattering geometries of a tetragonal superconductor can be described by the Raman response function with bare Raman scattering vertices (h = 1):

Imxrr(q-~O'co)=

4N(0)Re(\ coo~_-4;k12 IY~I21&12 > ' -

(31)

where N(0) is the density of states at the Fermi level for one spin orientation. On the other hand, screening effects are expected to strongly suppress scattering strength in the Alg (L=0) geometry, based upon the symmetry arguments delineated above. However, as discussed in more detail below, the observed Alg superconductingstate Raman response in the cuprates is not in general strongly suppressed, a fact which has motivated consideration of several other effects on the Alg-symmetry Raman scattering response, including resonance enhancements and multilayer effects (Krantz and Cardona 1994, 1995). 5.2.3. The influence of gap anisotropy It is important to note that the Tsuneto function in eq. (30) depends upon the square of the gap function, IAk[2, and hence the Raman scattering response in eq. (31) is sensitive only to the amplitude of the superconducting gap function, and not to its phase. Consequently, Raman scattering cannot distinguish, for example, between a

548

S.L. COOPER

d-wave pairing state, Ak = Ao(k2x- k2), and an anisotropic s-wave pairing state, such as Ak = A0 Ikx2 - k21, although Devereaux (1995) and Wu and Carbotte (1998) have suggested that electronic Raman scattering in the presence of impurities can be used to distinguish such pairing states. On the other hand, it is clear from eq. (31) that the Raman scattering response in the superconducting state can provide substantial information concerning the k-dependence of Ak, since eq. (31) involves an average of the T = 0 Tsuneto fimction (h = 1), ,~k = 4]AklZ/COV/co2 - 4]AkF, over the Fermi surface, weighted by the square of the Raman vertex, ]~/,k[2. Tlms, in a superconductor with an isotropic gap, Ak = constant, the Raman scattering response Imzry will have a polarization-independent "pair-breaking" peak at co=2A. On the other hand, in a superconductor with an anisotropic gap, Ak =f(k), the peak position and frequency dependence of the electronic scattering response depends upon the scattering geometry due to the weighting function I]/,kt2. 5.3. Analysis of the superconducting response in high-To superconductors 5.3.1. The dx2 y2 pairing state Devereaux et al. (1994a) and Devereaux and Einzel (1996a) first pointed out that in a tetragonal superconductor with a dx2 y2 pairing state, Ak = Ao(k 2 - k2), the asymptotic low-frequency behavior of I m z r r should exhibit an e)3 low-frequency dependence in the B i g electronic scattering response, but a linear-in-co dependence in the A l g and B2g symmetries (see table 2). This difference results rather generally from the fact that the T = 0 low-frequency density of states of a dx2 / superconductor is governed by the nodes, and hence varies linearly with energy. Consequently, one expects the lowf r e q u e n c y Bzg Raman response to vary linearly with frequency in a dx2- 9 superconductor, since the B2g vertex function is finite arotmd the nodes, while the Big-symmetry response varies as o)3 because the dx~_/ pairing and Big vertex functions vanish at the same points in k-space (see fig. 2). It has been pointed out that the presence of strong inelastic carrier scattering should significantly broaden and increase the energy of the peak in the Big response of a dx2 y2 superconductor, but should not strongly influence the 0)3 low-frequency dependence (Jiang and Carbotte 1995). As shown in fig. 16 (top), the calculated Raman cross-section for a dx2_y2 pairing state describes the observed superconducting-state electronic Raman scattering response of Bi2Sr2CaCu2Q Table 2 Tile asymptotic low-fiequency behavior of the electronic Raman scattering response function, Imzvv(co ~ 0), for a d,.2 y2 superconductor in different symmetries ~ Symmetry

Imxw(co ~ 0)

Big

Alg B2g After Devereaux et al. (1994a). Logarithmic factors have not been included.

~(ho)/2Ao)3 -ho/2A o ~h~o/2Ao

RAMAN SCATTERINGSTUDIESOF THE HIGH-T~CUPRATES

549

reasonably well (Devereaux et al. 1994a, Devereaux and Einzel 1996a). Subsequent measurements and analysis of the symmetry-dependent Raman scattering responses of other cuprates, including Laz-xSrxCuO4 (X.K. Chen et al. 1994a), T12Ba2CuO6 (Gasparov et al. 1997), and HgBazCa2Cu3Os (Sacuto et al. 2000) are also consistent with the scattering response of a dx~ y2 superconductor (see fig. 16). Notably, in the initial calculations of Devereaux et al. (1994a,b), the observed large anisotropy between the B2g and Ale pair-breaking energies was associated with Coulomb screening of the Ale component, which shifted this contribution down in energy from 2A to nearly A in the calculations. However, this effect is now known to be an artifact (Devereaux and Einzel 1996a,b) caused by ignoring higher-order Fermi-surface harmonics in the expansion of the Ale-symmetry Raman scattering vertex (eq. 16). More recent calculations of the Raman scattering cross-section in eq. (31) for a dx2_y2 superconductor, using more realistic approximations of the Raman vertex Ynk (X.K. Chen et al. 1994b, X.K. Chen and Irwin 1994, Branch and Carbotte 1995, Krantz et al. 1995, Krantz and Cardona 1995, Devereaux et al. 1996, Wenger and Kfill 1997, Strohm and Cardona 1997a), including approximations involving 2D tight-binding model calculations and 2D and 3D local-density approximation (LDA) band-structure calculations, do not reproduce the large measured differences in the peak positions of the Ale and Blg responses. Nevertheless, these calculations generally appear to support a dx2 y~ pairing state for the high-T~ superconductors, as they reproduce the data reasonably accurately in most other respects (Devereaux et al. 1996, Strohm and Cardona 1997a). It is clear from these calculations, however, that the fi'equency dependence of the low-frequency superconducting Raman response in different scattering geometries is a better probe of the pairing state than the anisotropy of the pair-breaking energy. Importantly, the large anisotropy in the measured Atg and Big peak energies remains unexplained - while calculations by Manske et al. (1997) suggest that screening of the Big Raman response is responsible for this difference, these calculations have been questioned [see Comment by Strohm, Munzar and Cardona (1998) and Reply by Manske et al. (1998)]. Indeed, the differences between the measured Ale and Big superconducting-state Raman responses may indicate that the Big response has a different origin than the Alg response: X.K. Chen et al. (1993) have suggested that the Big-symmetry peak in YBazCu307 in fact arises from spin fluctuations, based in part on the sensitivity of this response to doping. This issue clearly remains one of the outstanding questions facing future Raman scattering studies of high-Tc superconductors.

5.3.2. Effects of orthorhombic distortions While the initial calculations of Devereaux et al. (1994a,b) assumed a tetragonal superconductor with a dx2 y2 gap, orthorhombic distortions have consequences in a dx2 y~ superconductor that are expected to manifest themselves differently in the orthorhombic compounds YBazCu307 and Bi2Sr2CaCu2Os (Annett et al. 1996, Strohm and Cardona 1997a,b). In particular, in a tetragonal dx~_y2 cuprate superconductor, the Big-symmetry Raman vertex and the dx~ y2 gap function both have nodes in

550

S.L. COOPER

the [± 1, +1,0] directions, resulting in the characteristic 0)3 low-frequency Big Raman response discussed above. In YBa2Cu307, the orthorhombic distortion removes the mirror plane symmetry through (110), causing a shift of the Big Raman vertex nodes away from the [110] direction, and a consequent admixing of co and 0)3 low-energy scattering contributions in the Big geometry. By contrast, the orthorhombic unit cell in Bi2SrzCaCu208 is rotated by 45 ° relative to that of YBazCu307. Consequently, the orthorhombic distortion preserves the mirror plane symmetry along (110), and the superconducting-state Raman response in the Big scattering geometry is expected to maintain its pure ~)3 dependence. This difference may in part account for discrepancies observed between the Big spectra of YBazCu307 and Bi2Sr2CaCu208 (Devereaux and Einzel 1995), although Devereaux and Einzel have also pointed out that different Fano interaction effects between the optical phonons and low-frequency continua of YBa2Cu3Ov and BizSrzCaCu208 may contribute to the slightly different low-frequency spectral responses observed in these materials.

5.3.3. Multilayer effects One of the remaining puzzles related to the superconducting-state Raman response in the high-To cuprates is the observation of a substantial scattering response in the Alg geometry, which, as noted above, is in defiance of theoretical predictions that the Aig response should be strongly suppressed by Coulomb screening. Krantz and Cardona (1995) and Strohm and Cardona (1997a) have argued, by analogy with strong Alg scattering seen in multicomponent plasma systems, that effective mass fluctuations between different Fermi surface sheets in multilayer cuprate materials lead to the unscreened AIg scattering. A natural test of this supposition is to examine the superconducting Raman response of single-layer cuprates. Such measurements were made in single-layer T12BazCuO6 by Gasparov et al. (1997) and Kang et al. (1997), and it was found that the Alg scattering response is still strong in single-layer materials, although perhaps somewhat weaker than in multilayer materials (see middle plot, fig. 16). Recent studies of single-layer La2 xSrxCuO4 at optimal doping by X.K. Chen et al. ('1998a) also show no differences between the spectral response of single- and double-layer cuprates, further suggesting that multilayer effects on the normal- and superconductingstate Raman response of the cuprates are probably weak. Thus, the origin of the strong Alg Raman response remains an open question. Further study is needed to elucidate the role that strong Fermi-surface anisotropy, particularly that associated with the extreme in-plane/c-axis anisotropy, plays in the strong Ale scattering in the cuprates.

5.3.4. Other reported pairing states While the calculations of Devereaux et al. and others appear to show that many of the anomalous features observed in the superconducting-state electronic Raman scattering response of high-Tc cuprates can be explained assuming a dx2 y2 pairing state, there are several noteworthy exceptions to this interpretation.

RAMAN SCATTERINGSTUDIES OF THE HIGH-Tc CUPRATES

551

Stadlober et al. (1995) observed that the electron-doped system N d 2 - x C e x C u O 4 exhibits a redistribution of electronic scattering below To (~ 20K), but neither the pair-breaking energy nor the low-frequency co dependence of electronic scattering below To in Nd2_xCexCuO4 was found to be symmetry dependent. Moreover, the maximum pair-breaking energy inferred from these Raman scattering measurements, 2Amax~4.9kBTc, was much smaller than that estimated from studies of the holedoped cuprates, 2Z~max ~ 8 k B T c. Stadlober et al. (1995) concluded from these results that Nd2_xCexCuO4 is a nearly isotropic s-wave superconductor, corroborating the interpretation of temperature-dependent penetration depth measurements of this material (Wu et al. 1993). Kendziora et al. (1996) also reported that overdoped T12Ba2CuO6 and Bi2Sr2CaCu208 are isotropic s-wave superconductors, based upon their observation of a small pairbreaking energy, 2A ~ 3.9kBTc, and a small difference (~10%) between the Big- and B2gsymmetry "pair-breaking" peak energies in these materials. However, as mentioned above, the symmetry dependence of the "pair-breaking" energy is not a good measure of the pairing state in Raman scattering [concerning this point, see also Devereaux and Einzel (1995), as well as the Comment by Hewitt et al. (1997) and Reply by Kendziora and Onellion (1997)]. Instead, a better measure is the frequency dependence of low-energy electronic scattering in different scattering geometries, and recent measurements of the superconducting state Raman scattering response in overdoped T12Ba2CuO6 (Gasparov et al. 1998a,b) and Bi2Sr2CaCu208 (Blumberg et al. 1997b) reveal a co3 dependence in the Big geometry and a linear co dependence in the B2g geometry, characteristic of a dx2 y2 pairing state (see fig. 18). These results suggest strongly that the dx~_/ pairing state persists even into the overdoped phases of the Bi- and Tl-based cuprates. Finally, more complicated pairing states have also been inferred from superconductingstate Raman scattering data. For example, Krantz (1996) has considered the effects of interlayer pairing on the pairing state of YBa2Cu3OT, arguing that Raman scattering and tunneling data of this compound are consistent with the presence of Big- and Algsymmetry gaps associated with the even and odd Fermi surfaces, respectively (i.e., dx2 y2 +ias, a < 1). On the other hand, Annett et al. (1996) have argued against such even- and odd-parity pairing in bilayer materials based upon the fundamental similarity between the superconducting-state properties of one- and two-layer cuprates, and in particular upon the absence of evidence for a second transition in bilayer materials. Also, in Raman scattering measurements of optimally doped HgBa2CaCu206, Sacuto et al. (1997, 1998) [see also Correction to Sacuto et al. (1998) in Sacuto et al. (1999)] first reported a linear-in-co contribution to the low-frequency scattering response in both the Big and B2g geometries. As this material is tetragonal, and consequently immune to any mixing of co and 0)3 contributions due to orthorhombicity (see sect. 5.3.2), the authors concluded from this that the nodes in the gap function are shifted slightly away from the [-4-1,±1,0] directions in HgBa2CaCu206, and that the pairing function in this compound is consequently described by either a "g-wave" state [i.e., A2g symmetry -xy(x 2 _y2)] (Sacuto et al. 1997, 1998) or a d+ s state (Beal-Monod et al. 1997). However, in more recent measurements of underdoped HgBa2Ca2Cu3Os, in which sample surface

S.L. COOPER

552

a) B

100

-4

/~ ~" ~. f Tc=BOK .,: .¢"

lg

:~,/' M'~ ~., ,,~2 f ' ' v ,

/

E 033.5 /

..... . ,, ..,., .o. . . . . . .

50

0,5

1,0

......... 1

1,5

a)/0) (Big peak)

40

E

20

0.5

1 ,o / ~o (a2g ~o~)

1,5

Fig. 18. Symmetrydependenceof electronic Raman scattering in the superconductingstate of overdoped T12Ba2CuO6 (after Gasparovet al. 1998a,b), showing the o)3 low-frequencydependence of the Big scattering response and the linear-in-o)dependence in AI~ and Big.

problems were believed to be under better control, Sacuto et al. (2000) found frequency dependences of the Big, B2g, and Alg electronic Raman scattering intensities that were largely consistent with a dx2 _y2 pairing state. It is also perhaps appropriate to point out in this regard that the efficacy of Raman scattering for deducing subtle details related to the pairing state is limited by several factors, including the phase insensitivity of the technique (see sect. 5.2.3), the model dependence implicit in the analysis of the Raman response, and the inherent difficulties involved in extracting the intrinsic low-frequency scattering

RAMAN SCATTERINGSTUDIESOF THE HIGH-Tc CUPRATES

553

response due to material quality and extraneous scattering contributions. Therefore, subtle details associated with the pairing function, for example related to multilayer effects and precise nodal position, must ultimately be adjudicated by more sensitive (e.g., phasesensitive) techniques. 5.3.5. Relationship to the "normal"-state response Significantly, models of electronic Raman scattering in the superconducting state of high-To superconductors neither explain the anomalous normal-state electronic Raman continuum, nor rely upon this scattering to explain the development of a pair-breaking peak below To. Rather, peaks in the superconducting-state Raman scattering response are expected to develop near 2A in these models, even in the absence of normal-state scattering, due to the development of unscreened (i.e., L ~ 0) collective bound state and quasiparticle pair-breaking excitations below Tc (see, for example, Anderson 1958a,b, Bogoliubov 1958, Tsuneto 1960, Bardasis and Schrieffer 1961, Monien and Zawadowski 1990). However, Kang et al. (1996) reported experimental evidence for at least an indirect connection between the normal-state continuum and the superconducting-state "pair breaking" peak in T12BazCuO6, finding that both features exhibit a resonant enhancement associated with the 3.4 eV absorption band. They attribute this absorption band in T12Ba2CuO6 to a transition from the Fermi level to the upper Hubbard band (UHB), and associate the resonance profile of these features to the enhancement of the electronic and quasiparticle Raman scattering intensities by real interband transitions to the UHB. However, the nature of this connection between the normal- and superconducting-state Raman responses in TIzBa2CuO6 is not understood, nor has a similar connection been observed in other cuprates. 5.4. Electronic Raman scattering in high magnetic fields Until recently, Raman scattering studies of the high-Tc cuprates in high magnetic fields were primarily limited to studies of Raman-active phonons (Ruf et al. 1988, Podobedov et al. 1997, Loa et al. 1997) and crystal-field-phonon coupling (Ruf et al. 1992). Blumberg et al. (1997a) observed, however, that the Raman scattering "pair-breaking" peak in overdoped T12Ba2CuO6 is substantially suppressed by a magnetic field (see fig. 19). From the field and temperature dependencies of their Raman spectra, Blumberg et al. deduced that the upper critical field Hcz(T ) in overdoped TlzBazCuO6 has a conventional temperature dependence, in contrast to the anomalous temperature dependence of H~2(T ) inferred from magnetoresistance measurements (Dalichaouch et al. 1990, Almasan et al. 1992, Mackenzie et al. 1993, 1996, Osofsky et al. 1993). They further observed a nonlinear field dependence of the "pair-breaking" peak intensity, which they attributed to a renormalization of the quasiparticle density near the vortex cores. Interestingly, recent calculations of the Raman scattering intensity by Vekhter et al. (1999), which account for the Doppler energy shift of the extended quasiparticle states in the mixed phase of a d-wave superconductor, reveal several features that are qualitatively consistent with these measurements: (a) the Raman intensity below the average Doppler shift energy develops

554

S.L. COOPER

0 T-> ~x

T= 5K

14

o

~

= 1()

8.5T--~ ,~ 8

V

~¢'-

/ #

~/ 0

overdoped 50

T12BaaCuO 6

100

150

200

R a m a n S h i f t (cm "1) ,,

.

.

.

,

-

•

,

.

17K

*"

.

~

.

,

.

.

.

!

.

,

1

2 Magnetic

+

.

,

''i',."

'+

-~

-

o .... + - i o

i+ '+

3 Field

4

N 5

(T)

Fig. 19. (top) Magnetic field dependence of the superconducting-state electronic Raman scattering response in overdoped T12Ba2CuO6, showing a suppression of the "pair-breaking" peak with increasing magnetic field. (bottom) Field and temperature dependence of the "pair-breaking" peak intensity. (inset) Temperature dependence of the upper critical field, H*~z(T), obtained from fits to these data (after Blumberg et al. 1997a).

quartic and quadratic frequency dependences in the Big and B2g spectra, respectively; (b) the 2A peak intensity decreases linearly with magnetic field; and (c) the normalized integrated Raman intensity scales with V~. An interesting possible area of future Raman scattering investigation in high magnetic fields involves studies of the discrete quasiparticle spectrum inside the vortex cores of high-Tc superconductors. The energy spacing of these discrete states is governed by the

RAMAN SCATTERINGSTUDIESOF THE HIGH-ToCUPRATES

555

confinement energy associated with the vortex (Caroli, de Gennes and Matricon 1964), and thus should fall in an accessible frequency range for infrared and Raman spectroscopies in small-coherence-length superconductors such as the high-To superconductors (~0 ~ 10-20~Q. However, while quasiparticle pair excitations were reported by Karrai et al. (1992) using infrared measurements, there have as yet been no reports of vortex excitations observed with Raman scattering.

6. Conclusions There have been an enormous number of Raman scattering studies of the high-To superconductors in the past decade or so, but such is the richness of these remarkable materials that many veins of interesting physics have yet to be fully explored with this technique. First, Raman scattering in high magnetic fields is potentially interesting as a means of studying transitions between bound states in the vortex core. Second, Raman scattering in the underdoped cuprates is relatively untapped, and shows promise both as a means of investigating possible differences in the "pairing" and "phasing" temperatures in underdoped cuprates, and as a method for studying charge and spin domain structures that may exist in this phase regime. Raman scattering also lends itself to a number of techniques which may be particularly useful in the study of underdoped and other phases, including time-resolved, high-pressure, magnetic-field, and micro- and near-fieldRaman scattering techniques. For example, recent high-pressure Raman scattering studies of two-magnon spectra in lamellar cuprates have revealed interesting magnon-phonon interaction effects not accessible using other techniques (Struzhkin et al. 2000). Using Raman scattering to study high-To cuprate microstructures is also potentially interesting, not only as a means of spectroscopically investigating superconductivity in mesoscopic systems, but also as a versatile diagnostic tool for evaluating high-Tc devices (Bock et al. 1993). Finally, resonant X-ray Raman spectroscopy has recently evolved as a promising new probe of excitations in the cuprates, particularly in the insulating phase (Kuiper et al. 1998, Abbamonte et al. 1999). Indeed, given the remarkable, and often unexpected, past successes achieved in Raman scattering studies of high-To cuprates and other complex oxides, ultimately one can only speculate about the exciting directions that future Raman scattering studies of these materials will take.

Acknowledgements I would like to acknowledge the fruitful interactions I have had at Illinois with a number of people whose work in the area of Raman scattering from high-Tc superconductors or related topics contributed to this review: Pete Abbamonte, Girsh Blumberg, Don Ginsberg, Laura Greene, Moonsoo Kang, Marvin Karlow, Miles Klein, Art Kotz, Tony Leggett, Ran Liu, Paul Nyhus, Dmitri Reznik, Michael Rtibhausen, David Salamon, Myron Salamon, Francis Slakey, Charlie Slichter, Dale Van Harlingen, Doug Wake, and Ali Yazdani.

556

S.L. COOPER

I w o u l d like to particularly thank M i l e s K l e i n for his critical reading o f this m a n u s c r i p t prior to publication. Finally, I w o u l d like to thank m y wife, Laura Clower, and children, Duncan, Sarah, and Patrick, for their patience and support w h i l e I was writing this chapter. I also gratefully a c k n o w l e d g e support f r o m the D e p a r t m e n t o f E n e r g y u n d e r grant D E F G 0 2 - 9 6 E R 4 5 3 9 , and f r o m the S c i e n c e and T e c h n o l o g y Center for S u p e r c o n d u c t i v i t y under grant N S F D M R 9 1 - 2 0 0 0 0 .

References

Abbamonte, 13, C.A. Burns, E.D. Isaacs, EM. Platzman, L.L. Miller, S.-W. Cheong and M.V. Klein, 1999, Phys. Rev. Lett. 83, 860. Abrikosov, A.A., and L.A. Fal'kovskii, 1961, Sov. Phys. JETP 13, 179. Abrikosov, A.A., and L.A. Fal'kovskii, 1988, Physica C 156, 1. Abrikosov, A.A., and VM. Genkin, 1974, Sov. Phys. JETP 38, 417. Aeppli, G., S.M. Hayden, H.A. Mook, Z. Fisk, S.W. Cheong, D. Rytz, J.E Remeika, G.E Espinosa and A.S. Cooper, 1989, Phys. Rev. Lett. 62, 2052. Aharony, A., R.J. Birgeneau, A. Coniglio, M.A. Kastaler and H.E. Stanley, 1988, Phys. Rev. Lett. 60, 1330. Allen, EB., 1976, Phys. Rev. B 13, 1416. Almasan, C.C., M.C. de Andrade, Y. Dalichaouch, J.J. Neumeier, C.L. Seaman, M.B. Maple, R.R Guertin, M.V Kuric and J.C. Garland, 1992, Phys. Rev. Lett. 69, 3812. Andersen, O.K., A.I. Liechtenstein, C.O. Rodriguez, I.I. Mazin, O. Jepsen, MR Antropov, O. Gunnarsson and S. Gopalan, 1991, Physica C 185 189, 147. Andersen, O.K., O. Jepsen, A.I. Liechtenstein and I.I. Mazin, 1994, Phys. Rev. B 49, 4145. Anderson, RV~, 1958a, Phys. Rev. 110, 827. Anderson, EV~, 1958b, Phys. Rev. 112, 1900. Anderson, EW., 1987, Science 235, 1196. Anderson, EW., 1997, The Theory of Superconductivity in High T c Cuprates (Princeton University Press, Princeton, NJ). Annett, J.E, N.D. Goldenf~ld and A.J. Lcggett, 1996, in: Physical Properties of High Temperature Superconductors, Vol. V, ed. D.M. Ginsberg (World Scientific, Singapore)p. 375. Aronson, M.C., S.B. Dierker, B.W Dennis, S.W Cheong and Z. Fisk, 1991, Phys. Rev. B 44, 4657.

Bacci, S.B., and E.R. Gagliano, 1991, Phys. Rev. B 43, 6224. Balseiro, C.A., and L.M. Falicov, 1980, Phys. Rev. Lett. 45, 662. Bardasis, A., and J.R. Schrieffer, 1961, Phys. Rev. 121, 1050. Basu, S.,and A. Singh, 1997, Phys. Rev. B 55, 12338. Batlogg, B., and VJ. Emery, 1996, Nature 382, 20. Batlogg, B., H.Y. Hwang, H. Takagi, R.J. Cava, H.L. Kao and J. Kwo, 1994, Physica C 235-240, 130. Bazhenov, A.V., A.V Gorbunov, N.V Klassen, S.E Kodakov, I.V Kukushkin, V.D. Kalakovskii, O.V Misochko, V.B. Timofeev, L.I. Chernyshova and B.N. Shepel, 1987, Pis'ma Zh. Eksp. Teor. Fiz. 46, 35 [JETP Lett. 46, 32]. Beal-Monod, M.T., J.B. Bieri and K. Maki, 1997, Europhys. Lett. 40, 201. Blumberg, G., R. Liu, M.V Klein, W.C. Lee, D.M. Ginsberg, C. Gu, B.W. Veal and B. Dabrowski, 1994, Phys. Rev. B 49, 13295. Blumberg, G., R Abbamonte, M.V Klein, W.C. Lee, D.M. Ginsberg, L.L. Miller and A. Zibold, 1996, Phys. Rev. B 53, Rl1930. Blumberg, G., M. Kang and M.V Klein, 1997a, Phys. Rev. Lett. 78, 2461. Blumberg, G., M. Kang, M.V. Klein, K. Kadowaki and C. Kendziora, 1997b, Science 278, 1427. Bhtmberg, G., M.V Klein and S.-W. Cheong, 1998a, Phys. Rev. Lett. 80, 564. Blumberg, G., M.V. Klein, K. Kadowaki, C. Kendziora, R Guptasarma and D. Hinks, 1998b, J. Phys. Chem. Solids 59, 1932. Bock, A., J.C. Fretwurst and U. Merkt, 1993, J. Phys. Chem. Solids 54, 1343. Bogoliubov, N.N., 1958, Nuovo Cim. 7, 6. Bozovic, I., D. Klrillov, A. Kapitulnik, K. Char, M.R. Hahn, M.R. Beasley, T.H. Geballe, Y.H. Kim and A.J. Heegcr, 1987, Phys. Rev. Lett. 59, 2219.

RAMAN SCATTERING STUDIES OF THE HIGH-T~ CUPRATES Bozovic, I., J.H. Kim, J.S. Harris, C.B. Eom, J.M. Phillips and J.T. Chetmg, 1994, Phys. Rev. Lett. 73, 1436. Branch, D., and J.R Carbotte, 1995, Phys. Rev. B 52, 603. Brawner, D.A., and H.R. Ott, 1994, Phys. Rcv. B 50, 6530. Brenig, W, and H. Monien, 1992, Solid State Commun. 83, 1009. Browne, D.A., and K. Levin, 1983, Phys. Rev. B 28, 4029. Canali, C.M., and S.M. Girvin, 1992, Plays. Rev. B 45, 7127. Cardona, M., 1999, Physica C 318, 30. Caroli, C., RG. de Gennes and J. Matricon, 1964, Phys. Lett. 9, 307. Cerdeira, E, N. Mestres and M. Cardona, 1984, Phys. Rev. B 29, 3737. Chandrasekhar, M., M. Cardona and E.O. Kane, 1977, Phys. Rev. B 16, 3579. Chaudhari, E, and S.-Y. Lin, 1994, Phys. Rev. Lett. 72, 1084. Chen, C.H., S.-W. Cheong and A.S. Cooper, 1993, Plays. Rev. Lett. 71, 2461. Chen, C.T., L.H. Tjeng, J. Kwo, H.L. Kao, P. Rudolf, E Sette and R.M. Flemming, 1992, Phys. Rev. Lett. 68, 2543. Chen, X.K., and J.C. Irwin, 1994, Physica 235-240, 1087. Chen, X.K., E. Altendorf, J.C. hwin, R. Liang and W.N. Hardy, 1993, Phys. Rev. B 48, 10530. Chert, X.K., J.C. Irwin, H.J. Trodahl, T. Kimura and K. Kishio, 1994a, Phys. Rev. Lett. 73, 3290. Chen, X.K., J.C. Irwin, R. Liang and W.N. Hardy, 1994b, Physica C 227, 113. Chen, X.K., J.G. Naeini, K.C. Hewitt, J.C. irwin, R. Liang and W.N. Hardy, 1997, Phys. Re,/. B 56, R513. Chert, X.K., J.C. Irwin, H.J. Trodahl, M. Okuya, "12 Kimura and K. Kishio, 1998a, Physica C 295, 80. Chen, X.K., J.G. Naeini, J.C. Irwin, R. Liang and W.N. Hardy, 1998b, J. Phys. Chem. Solids 59, 1968. Chen, Y., E Wilczek, E. Witten and B. Halperin, 1989, Int. J. Mod. Phys. B 3, 1001. Chubukov, A.V, and D.M. Frenkel, 1995, Phys. Rev. Lett. 74, 3057. Contreras, G., A.K. Sood and M. Cardona, 1985, Phys. Rev. B 32, 930. Cooper, S.L., and K.E. Gray, 1994, in: Physical Properties of High Temperature Superconductors,

557

Vol. IV,, ed. D.M. Ginsberg (World Scientific, Singapore) p. 61. Cooper, S.L., and M.V Klein, 1990, Comm. Condens. Matter Phys. 15, 19. Cooper, S.L., M.V Klein, B.G. Pazol, J.E Rice and D.M. Ginsbcrg, 1988a, Plays. Rev. B 37, 5920. Cooper, S.L., E Slakey, M.V Klein, J.R Rice, E.D. Bukowski and D.M. Ginsberg, 1988b, Phys. Rev. B 38, 11934. Cooper, S.L., G.A. Thomas, A.J. Millis, RE. Sulewski, J. Orenstein, D.H. Rapkine, S.-W. Cheong and EL. Trevor, 1990, Phys. Rcv. B 42, 10785. Cooper, S.L., R Nyhus, D. Reznik, M.V. Klein, WC. Lee, D.M. Ginsberg, B.V~ Veal, A.P. Paulikas and B. Dabrowski, 1993a, Phys. Rev. Lett. 70, 1533. Cooper, S.L., D. Reznik, A.L. Kotz, M.A. Karlow, R. Liu, M.V Klein, W.C. Lee, J. Giapintzakis, D.M. Ginsberg, B.W. Veal and A.P. Paulikas, 1993b, Phys. Rev. B 47, 8233. Cuden, C.B., t976, Phys. Rev. B 13, 1993, Dagotto, E., 1994, Rev. Mod. Phys. 66, 763. Dalichaouch, Y., B.W. Lee, C.L. Seaman, J.T. Markcrt and M.B. Maple, 1990, Phys. Rev. Lett. 64, 599. Devereaux, ~ER, 1992, Phys. Rev. B 45, 12965. Devereaux, T.E, 1993, Phys. Rev. B 47, 5230. Devereaux, T.E, 1995, Phys. Rev. Lett. 74, 4313. Devereaux, ZE, and D. Einzel, 1995, Phys. Rev. B 51, 16336. Devereaux, T.R, and D. Einzel, 1996a, Phys. Rev. B 54, 15547(E). Devereaux, T.R, and D. Einzel, 1996b, Phys. Rev. B 54, 15548(E). Devereaux, T.P., D. Einzel, B. Stadlober, R. Hackl, D.H. Leach and J.J. Neumeier, 1994a, Phys. Rev. Lett. 72, 396. Devereaux, T.P., D. Einzel, B. Stadlober and R. Hackl, 1994b, Phys. Rev. Lett. 72, 3291. Devereaux, T.R, A. Virosztek and A. Zawadowski, 1996, Plays. Rev. B 54, 12523. Dierker, S.B., M.V Klein, G.W. Webb and Z. Fisk, 1983, Phys. Rev. Lett. 50, 853. Ding, H., J,C. Campuzano, K. Gofron, G. Gu, R. Liu, B.W. Veal and G. Jennings, 1994, Phys. Rev. B 50, 1333. Ding, H., T. Yokoya, J.C. Campuzano, T. Takahashi, M. Randeria, M.R. Norman, T. Mochiku, K. Kadowald and J. Giapintzakis, 1996, Nature 382, 51. Donovan, S., J. Kircher, J. Karpinski, E. Kaldis and M. Cardona, 1995, J. Supercond. 8, 417.

558

S.L. COOPER

Dufour, R, S. Jandl, C. Thomsen, M. Cardona, B.M. Wanklyn and C. Changkang, 1995, Phys. Rev. B 51, 1053. Elliot, R.J., and M.E Thorpe, 1969, J. Phys. C 2, 1630. Elliot, R.J., M.F. Thorpe, G.E Inbusch, R. Loudon and J.B. Parkinson, 1968, Phys. Rev. Lett. 21, 147. Emery, VJ., and S.A. Kivelson, 1995, Phys. Rev. Lett. 74, 3253. Emery, V.J., S.A. Klvelson and O. Zachar, 1997, Phys. Rev. B 56, 6120. Fatck, J.R, A. Levy, M.A. Kastner and R.J. Birgeneau, 1992, Phys. Rev. Lett. 69, 1109. Fal'kovskii, L.A., and S. Klama, 1990, Physica C 172, 242. Fleury, RA., and H.J. Guggenheim, 1970, Phys. Rev. Lett. 24, 1346. Fleury, RA., and R. Loudon, 1968, Phys. Rev. 166, 514. Fong, H.E, B. Keimer, D. Reznik, D.L. Milius and I.A. Aksay, 1995, Phys. Rev. Lett. 75, 316. Fong, H.E, B. Keimer, P.W. Anderson, D. Reznik, E Dogan and I.A. Aksay, 1996, Phys. Rev. B 54, 6708. Freitas, RJ., and R.R.R Singh, 2000, Phys. Rev. B 62, 5525. Gasparov, L.M, P. Lemmens, M. Brinkmann, N.N. Kolesnikov and G. Gfintherodt, 1997, Phys. Rev. B 55, 1223. Gasparov, L.V, R Lemmens, N.N. Kolesnikov and G. G/Jntherodt, 1998a, Physica B 244, 54. Gasparov, L.V, P. Lemmens, N.N. Kolesnikov and G. GiLntherodt, 1998b, Phys. Rev. B 58, 11753. Ghiron, K., M.B. Salamon, B.W. Veal, A.R Paulikas and J.W. Downey, 1992, Phys. Rev. B 46, 5837. Haas, S., E. Dagotto, J. Riera, R. Merlin and E Nori, 1994, J. Appl. Phys. 75, 6340. Hackl, R., and R. Kaiser, 1988, J. Phys. C 21, L453. Hackl, R., R. Kaiser and S. Schicktanz, 1983, J. Phys. C 16, 1729. Hackl, R., W. Gl/iser, P. M/iller, D. Einzel and K. Andres, 1988, Phys. Rcv. B 38 7133. Hackl, R., R. Kaiser and W. Glhser, 1989, Physica C 162 164, 431. Hackl, R., M. Opel, P. Mtiller, G. Krug, B. Stadlober, R. Nemetschek, H. Berger and L. Forro, 1996, J. Low Temp. Phys. 105, 733. Hayden, S.M., G. Aeppli, H.A. Mook, T.G. Perring, T.E. Mason, S.-W. Cheong and Z. Fisk, 1996, Plays. Rev. Lett. 76, 1344.

Hayes, W., and R. Loudon, 1978, Scattering of Lig~lt by Crystals (Wiley, New York). Hewitt, K.C., T.R Devereaux, X.K. Chen, X.-Z. Wang, J.G. Naeini, A.E. Curzon and J.C. Irwin, 1997, Phys. Rev. Lett. 78, 489t. Hewitt, K.C., N.L. Wang, J.C. Irwin, D.M. Pooke, A.E. Pantoja and H.J. Trodahl, 1999, Phys. Rev: B 60, R9943. Heyen, E.T., S.N. Rashkeev, I.I. Mazin, O.K. Andersen, R. Liu, M. Cardona and O. Jepsen, 1990, Phys. Rev. Lett. 65, 3048. Heyen, E.T., R. Wegerer, E. Sch6nherr and M. Cardona, 1991, Phys. Rev. B 44, 10195. Heyen, E.T., J. Klrcher and M. Cardona, 1992, Phys. Rev. B 45, 3037. Hoffmann, A., E Lemmens, G. Gtintherodt, V. Thomas and K. Winzer, 1994, Physica C 235-240, 1897. Homes, C.C., T. Timusk, R. Liang, D.A. Bonn and W.N. Hardy, 1993, Phys. Rev. Lett. 71, 1645. iliev, M., C. Thomsen, V. Hadjiev and M. Cardona, 1993, Phys. Rev. B 47, 12341. Ipatova, I.P., A.V Subashiev and V.A. Voitenko, 1981, Solid State Commun. 37, 893. ltai, K., 1992, Phys. Rev. B 45, 707. Ito, T., Y. Nakamura, H. Takagi and S. Uchida, 1991, Physica C 185-189, 1267. Ito, T., K. Takenaka and S. Uchida, 1993, Phys. Rev. Lett. 70, 3995. Jandl, S., P. Dufour, T. Ruf, M. Cardona, V. Nekvasil, C. Chen and B.M. Wanklyn, 1995, Phys. Rev. B 52, 15558. Jiang, C., mad J.P. Carbotte, 1995, Solid State Commnn. 95, 643. Jongh, LJ., and R. Block, 1975, Physica B+C 79, 568. Kampf, A.P., and W. Brenig, 1992, Z. Phys. B 89, 313. Kang, M., G. Blumberg, M.V. Klein and N.N. Kolesnikov, 1996, Plays. Rev. Lett. 77, 4434. Kang, M., G. Blmnberg, M.V. Klein and N.N. Kolesnikov, 1997, Phys. Rev. B 56, R11427. Karrai, K., E.-J. Choi, E Dunmore, S. Liu, H.D. Drew, Q. Li, D.B. Fenner, Y.D. Zhu and E-C. Zhang, 1992, Phys. Rev. Lett. 69, 152. Katsufuji, T., and Y. Tokura, 1994, Phys. Rcv. B 49, 4372. Katsufuji, T., Y. Tokm'a, T. Ido and S. Uchida, 1993, Phys. Rev. B 48, 16131. Kendziora, C., and M. Onellion, 1997, Phys. Rev. Lett. 78, 4892.

RAMAN SCATTERING STUDIES OF THE HIGH-Tc CUPRATES Kendziora, C., and A. Rosenberg, 1995, Phys. Rev. B 52, R9867. Kendziora, C., R. Kelley and M. Onellion, 1996, Phys. Rev. Lett. 77, 727. Khveshehenko, D.V, and RB. Wiegmann, 1994, Phys. Rev. Lett. 73, 500. Kirillov, D., I. Bozovic, T.H. Geballe, A. Kapitulnik and D.B. Mitzi, 1988, Phys. Rev. B 38, 11955. Klein, M.V., 1975, in: Light Scattering in Solids, Vol. I, ed. M. Cardona, Vol. 8 of Topics in Applied Physics (Springer, Berlin) p. 147. Kleha, M.V, and S.B. Dierker, 1984, Plays. Rev. B 29, 4976. Klein, M.'~, S.L. Cooper, E Slakey, LR Rice, E.D. Bukowski and D.M. Ginsberg, t989, in: Strong Con'elations and Superconductivity, eds H. Fukuyama, S. Maekawa and A.R Malozemoff (Springer, Berlin) p. 226. Knoll, R, C. Thomsen, M. Cardona and E Murugaraj, 1990, Phys. Rev. B 42, 4842. Kostur, VN., 1992, Z. Phys. B 89, 149. Kosztin, J., and A. Zawadowski, 1991, Solid State Commun. 78, 1029. Kouznetsov, K.A., A.G. Sun, B. Chen, A.S. Katz, S.R. Bahcall, J. Clarke, R.C. Dynes, D.A. Gajewski, S.H. Han, M.B. Maple, J. Giapintzakis, J.-T. Kim and D.M. Ginsberg, 1997, Phys. Rev. Lett. 79, 3050. Krantz, M.C., 1996, Phys. Rev. B 54, 1334. Krantz, M.C., and M. Cardona, 1994, Phys. Rev. Lett. 72, 3290. Krantz, M.C., and M. Cardona, 1995, J. Low Temp. Phys. 99, 205. Krantz, M.C., H.J. Rosen, J.Y.T. Wei and D.E. Morris, 1989, Phys. Rev. B 40, 2635. Krantz, M.C., 1.1. Mazin, D.H. Leach, W.Y. Lee and M. Cardona, 1995, Phys. Rev. B 51, 5949. Krol, D.M., M. Stavola, L.E Schneemeyer, J.V Waszczak and S.A. Sunshine, 1989, J. Opt. Soc. Am. B 6, 448. Kuiper, R, J.-H. Guo, C. Sfithe, L.-C. Duda, J. Nordgren, J.J.M. Pothuizen, EM.E de Groot and G.A. Sawatzky, 1998, Phys. Rev. Lett. 80, 5204. Kumar, N., and A.M. Jayannavar, 1992, Phys. Rev. B 45, 5001. Lee, J.D., and B.I. Min, 1996, Solid State Commun. 99, 627. Leggett, A.J., 1998, J. Phys. Chem. Solids 59, 1729. Lei, X.L., J.L. Birman and C.S. Ting, 1985, Physica B 135, 486. Littlewood, RB., and C.M. Varma, 1981, Phys. Rev. Lett. 47, 811.

559

Littlewood, EB., and C.M. "?anna, 1982, Phys. Rev. B 26, 4883. Litvinchuk, A.R, C. Thomsen and M. Cardona, 1992, Solid State Commun. 83, 343. Liu, H.L., G. Blumberg, M.V Klein, P. Guptasm'rna and D.G. Hinks, 1999, Phys. Rev. Lett. 82, 3524. Liu, R., D. Salamon, M.V. Klein, S.L. Cooper, W.C. Lee, S.-W. Cheong and D.M. Ginsberg, 1993a, Phys. Rev. Lett. 71, 3709. Liu, R., M.V Klein, D. Salanaon, S.L. Cooper, VV~C. Lee, S.-W. Cheong and D.M. Ginsberg, 1993b, J. Phys. Chem. Solids 54, 1347. Loa, I., J. Hofmann and C. Thomsen, 1997, Phys. Status Solidi b 201, R3. Loeser, A.G., Z.-X. Shen, D.S. Dessau, D.S. Marshall, C.-H. Park, R Fournier and A. Kapitulnik, 1996, Science 273, 325. Loram, J.W., K.A. Mirza, J.R. Cooper and W.Y. Liang, I993, Phys. Re,/. Lett. 71, 1740. Lorenzana, J., and G.A. Sawatzky, 1995a, Phys. Rev. Lett. 74, 1867. Lorenzana, J., and G.A. Sawatzky, 1995b, Phys. Rev. B 52, 9576. Lyons, K.B., S.H. Liou, M. Hong, H.S. Chen, J. Kwo and T.J. Negran, 1987, Phys. Re,/. B 36, 5592. Lyons, K.B., RA. Fleury, J.R Remeika, A.S. Cooper and T.J. Negran, 1988a, Phys. Rev. B 37, 2353. Lyons, K.B., EA. Fleury, L.E Schneemeyer and J.V Waszczak, 1988b, Phys. Rev. Lett. 60, 732. Lyons, K.B., EA. F l e w , R.R.R Singh and RE. Sulewski, 199l, in: Dynamics of Magnetic Fluctuations in High Temperatm'e Superconductors, eds G. Reiter, R Horsch and G.C. Psaltakis (Plenum, New York) p. 159. Mackenzie, A.E, S.R. Julian, G.G. Lonzarich, A. Carrington, S.D. Hughes, R.S. Liu and D.C. Sinclair, 1993, Phys. Rev. Lett. 71, 1238. Mackenzie, A.R, S.R. Julian, D.C. Sinclair and C.T. Lin, 1996, Phys. Rev. B 53, 5848. Maksimov, A.A., and I.I. Tartakovsldi, 1994, J. Supercond. 7, 439. Manske, D., C.T. Rieck, R. Das Sharma, A. Bock and D. Fay, 1997, Phys. Rev. B 56, R2940. Manske, D., A. Bock, C.T. Rieck and D. Fay, 1998, Phys. Rev. B 58, 8841. Marshall, D.S., D.S. Dessau, A.G. Loeser, C.H. Park, A.Y. Matsuvra, J.N. Eckstein, I. Bozovic, E Fournier, A. Kapitutnik, W.E. Spicer and Z.X. Shen, 1996, Phys. Rev. Lett. 76, 4841. Micnas, R., J. Ramlinger, S. Robaszkiewicz and S. Tabor, 1988, Phys. Rev. B 37, 9410.

560

S.L. COOPER

Micnas, R., J. Ranninger and S. Robaszkiewicz, 1990, Rev. Mod. Phys. 62, 113. Mihailovic, D., K.E McCarty and D.S. Ginley, 1993, Phys. Rev. B 47, 8910. Monien, H., and A. Zawadowski, 1990, Phys. Rev. B 41, 8798. Moriya, T., Y. Takahashi and K. Ueda, 1990, J. Phys. Soc. Japan 52, 2905. Morr, D.K., and A.V Chubukov, 1997, Phys. Rev. B 56, 9134. Naeini, J.G., X.K. Chen, K.C. Hewitt, J.C. Irwin, T.R Devereaux, M. Okuya, T. Kimura and K. Kishio, 1998, Phys. Rev. B 57, Rl1077. Naeini, J.G., X.K. Chen, J.C. Irwin, M. Okuya, T. Kimnra and K. Kishio, 1999, Phys. Rev. B 59, 9642. Nagansa, N., and RA. Lee, 1990, Phys. Rev. Lett. 64, 2450. Nagaosa, N., and EA. Lee, 1991, Phys. Rev. B 43, 1233. Nemetschek, R., O.V Misochko, B. Stadlober and R. Hackl, 1993, Phys. Rev. B 47, 3450. Nemetschek, R., M. Opel, C. Hoffmann, RE Mfiller, R. Hackl, H. Berger, L. Forro, A. Erb and E. Walker, 1997, Phys. Rev. Lett. 78, 4837. Nori, E, E.R. Gagliano and S.B. Bacci, 1992, Phys. Rev. Lett. 68, 240. Nori, E, R. Merlin, S. Haas, A.W. Sandvik and E. Dagotto, 1995, Phys. Rev. Lett. 75, 553. Nyhus, R, M.A. Karlow, S.L. Cooper, B.W Veal and A.R Paulikas, 1994, Phys. Re,/. B 50, 13898. Nyhns, E, S. Yoon, M. Kauffinan, S.L. Cooper, Z. Fisk and J. Sarrao, 1997, Phys. Rev. B 56, 2717. Ohta, Y., T. Tohyama and S. Maekawa, 1991, Phys. Rev. Lett. 66, 1228. Opel, M., R. Nemetschek, C. Hoffmann, R. Philipp, RE Miiller, R. Hackl, I. Tutto, A. Erb, B. Revaz, E. Walker, H. Berger and L. Forro, 2000, Phys. Rev. B 61, 9752. Orenstein, J., G.A. Thomas, A.J. Millis, S.L. Cooper, D.H. Rapkine, T. Timusk, L.E Schneemeyer and LV Waszczak, 1990, Phys. Rev. B 42, 6342. Osofsky, M.S., R.J. Soulen, S.A. Wolf, J.M. Broto, H. Rakoto, J.C. Ousset, G. Coffe, S. Askenazy, R Pari, I. Bozovic, J.N. Eckstein and G.E Virshup, 1993, Phys. Rev. Lett. 71, 2315. Parkinson, J.B., 1969, J. Phys. C 2, 2012. Pennington, C.H., D.J. Durand, C.R Slichter, J.R Rice, E.D. Bukowski and D.M. Ginsberg, 1989, Phys. Rev. B 39, 274.

Perkins, J.D., J.M. Graybeal, M.A. Kastner, R.J. Birgeneau, J.E Falck and M. Greven, 1993, Phys. Rev. Lett. 71, 1621. Pines, D., 1963, Elementary Excitations in Solids (Addison-Wesley, Reading, MA). Pines, D., 1997, Physica C 282-287, 273. Plakida, N.M., VS. Oudovenko, R Horsch and A.I. Liechtenstein, 1997, Phys. Rev. B 55, R11997. Podobedov, V.B., J.E Rice, A. Weber and H.D. Drew, 1997, J. Supercond. 10, 205. Prelovsek, R, and J. Jaldic, 1996, Phys. Rev. B 53, 15095. Puchkov, A.M, D.N. Basov and T. Timusk, 1996, J. Phys.: Condens. Matter 8, 10049. Quilty, J.W., H.J. Trodahl and D.M. Pooke, 1998, Phys. Rev. B 57, Rl1097. Quilty, J.W, S. Tajima, S. Adachi and A. Yamanaka, 2001, Plays. Rev. B 63, 100508(11). Ranninger, J., and J.M. Robin, 1996, Phys. Rev. B 53, R11961. Rashkeev, S.N., and G. Wendin, 1993, Phys. Rev. B 47, 11603. Reznik, D., M.V Klein, W.C. Lee, D.M. Ginsberg and S.-W. Cheong, 1992, Phys. Rev. B 46, 11725. Reznik, D., S.L. Cooper, M.V Klein, V~C. Lee, D.M. Ginsberg, A.A. Maksimov, A.V Puchkov, I.I. Tartakovskii and S.-W. Cheong, 1993, Phys. Rev. B 48, 7624. Rojo, A.G., and K. Levin, 1993, Phys. Rev. B 48, 16861. Rossat-Mignod, J., L.E Regnanlt, C. Vettier, R Burlet, J.Y. Henry anrt G. Lapertot, 1991, Physica B 169, 58. Rfibhausen, M., N. Dieckmann, A. Bock, U. Merkt, W. Widder and H.E Braun, 1996, Phys. Rev. B 53, 8619. Rfibhausen, M., C.T. Rieck, N. Dieckmann, K.O. Subke, A. Bock and U. Merki, 1997, Phys. Rev. B 56, 14797. Rfibhansen, M., R Guptasarma, D.G. Hinks and M.V Klein, 1998, Phys. Rev. B 58, 3462. Riibhausen, M., O.A. Harmnerstein, A. Bock, U. Merkt, C.T. Rieck, R Guptasarma, D.G. Hinks and M.V Klein, 1999, Phys. Rev. Lett. 82, 5349. Ruf, T., C. Thomsen, R. Liu and M. Cardona, 1988, Phys. Rev. B 38, 11985. Ruf, T., E.T. Heyen, M. Cardona, J. Mesot and A. Furrer, 1992, Phys. Rev. B 46, 11792. Sacuto, A., R. Combescot, N. Bontemps, E Monod, V. Viallet and D. Colson, 1997, Europhys. Lett. 39, 207.

RAMAN SCATTERING STUDIES OF THE HIGH-To CUPRATES Sacuto, A., R. Combescot, N. Bontemps, C.A. Mtiller, V Viallet and D. Colson, 1998, Phys. Rev. B 58, 11721. Sacuto, A., R. Combescot, N. Bontemps, C.A. Miiller, V Viallet and D. Colson, 1999, Phys. Rev. B 59, 11605. Sacuto, A., J. Cayssol, P Monod and D. Colson, 2000, Phys. Rev. B 61, 7122. Salamon, D., R. Lin, M.V Klein, M.A. Karlow, S.L. Cooper, S.-W Cheong, W.C. Lee and D.M. Ginsberg, 1995, Phys. Rev. B 51, 6617. Salamon, D., E Abbamonte, R. Liu, M.V Klein, WC. Lee, D.M. Ginsberg, I.I. Tartakovskii and B.W. Veal, 1996, Phys. Rev. B 53, 886. Sandvik, A.W., S. Capponi, D. Poilblanc and E. Dagotto, 1998, Phys. Rev. B 57, 8478. Sanjurjo, J.A., G.B. Martins, EG. Pagliuso, E. Granado, I. Torriani, C. Rettori, S. Oseroff aud Z. Fisk, 1995, Phys. Rev. B 51, 1185. Scalapino, D.J., 1995, Phys. Rep. 250, 330. Sch6nfeld, F., A.R Kampf and E. Mtiller-Hartmann, 1997, Z. Phys. B 102, 25. Shastry, B.S., and B.I. Shraiman, 1990, Phys. Rev. Lett. 65, 1068. Shastry, B.S., and B.I. Shraiman, 199l, Int. J. Mod. Phys. B 5, 365. Shen, Z.-X., D.S. Dessau, B.O. Wells, D.M. King, W.E. Spicer, A.J. Arko, D.S. Marshall, L.W. Lombardo, A. Kapitulnik, R Dickinson, S. Doniach, J. DiCarlo, A.G. Loeser and C.-H. Park, 1993, Phys. Rev. Lett. 70, 1553. Simon, M.E., A.A. Aligia, C.D. Batista, E.R. Gagliano and E Lema, t996, Phys. Rev. B 54, R3780. Singh, R.R.R, PA. Fleury, K.B. Lyons and RE. Sulewski, 1989, Phys. Rev. Lett. 62, 2736. Slakey, E, M.V Klein, E.D. Bukowski and D.M. Ginsberg, 1990a, Phys. Rev. B 41, 2109. Slakey, F., M.V. Klein, J.R Rice and D.M. Ginsberg, 1990b, Phys. Rev. B 42, 2643. Slakey, F., M.V. Klein, J.R Rice and D.M. Ginsberg, 1991, Phys. Rev. B 43, 3764. Sokol, A., and D. Pines, 1993, Plays. Rev. Lett. 71, 2813. Sooryakumar, R., and M.V Klein, 1980, Phys. Rev. Lett. 45, 660. Sooryakumar, R., and M.V Klein, 1981, Phys. Rev. B 23, 3213. Stadlober, B., G. Krug, R. Nemetschek and R. Hackl, 1995, Phys. Rev. Lett. 74, 4911. Staufer, T., R. Nemetschek, R. Hackl, E Mtiller and H. Veith, 1992, Phys. Rev. Lett. 68, 1069.

561

Strohm, T., and M. Cardona, 1997a, Phys. Rev. B 55, 12725. Strohm, T, and M. Cardona, 1997b, Solid State Commun. 104, 233. Strohm, T., D. Munzar and M. Cardona, 1998, Phys. Rev. B 58, 8839. Struzhkin, VV, A.E Goncharov, H.K. Mao, R.J. Hemley, S.W. Moore, J.M. Graybeal, J. Sarrao and Z. Fisk, 2000, Phys. Rev. B 62, 3895. Sugai, S., 1989, Phys. Rev. B 39, 4306. Sugai, S., andY. Hidaka, 1991, Phys. Rev. B 44, 809. Sugai, S., H. Takagi, S. Uchida and S. Tanaka, t988, Jpn. J. Appl. Phys. 27, L1290. Sugai, S., Y. Enomoto and T. Murakami, 1989, Solid State Commun. 72, 1193. Sugai, S., M. Sato, T. Kobayashi, J. Akimitsu, T. Ho, H. Takagi, S, Uchida, S. Hosoya, T. Kajitani and T. Fukuda, 1990a, Phys. Rev. B 42, 1045. Sugai, S., M. Sato, T. Ito, T. Ido, H. Takagi, S. Uchida, T. Kobayashi, J. Akimitsu, Y. Hidaka, T. Murakami, S. Hosoya, T. Kajitani and T. Fukuda, 1990b, Physica B 165-166, 1263. Sulewski, P.E., PA. Fleury, K.B. Lyons, S.-W. Cheong and Z. Fisk, 1990, Phys. Rev. B 41, 225. Sulewski, P.E., PA. Fleury, K.B. Lyons and S.-W. Cheong, 1991, Phys. Rev. Lett. 67, 3864. Sun, A.G., D.A. Gajewski, M.B. Maple and R.C. Dynes, 1994, Phys. Rev. Lett. 72, 2267. Tajima, S., Y. Mizuo, M. Yoshida, T. Wada, Y. Yaegashi, T. Takata, M. Kagiya, N. Koshizuka, H. Uamauchi, K. Tamasaku and S. Uchida, 1993, Phys. Rev. B 47, 12126. Tanner, D.B., and T. Timusk, 1992, in: Physical Properties of High Temperature Superconductors, Vol. III, ed, D.M. Ginsberg (World Scientific, Singapore) p, 61. Thomsen, C., 1991, in: Light Scattering in Solids. Vol. VI, eds M. Cardona and G. G/,intherodt, Vol. 68 of Topics in Applied Physics (Springer, Berlin) p. 285. Thomsen, C., and M. Cardona, 1989, in: Physical Properties of High Temperature Superconductors, Vol. I, ed. D.M. Ginsberg (World Scientific, Singapore) p. 409. Thomsen, C., M. Cardona, B. Gegenheimer, R. Lin and A. Simon, 1988, Phys. Rev. B 37, 9860. Tinkham, M., 1975, Introduction to Superconductivity (McGraw-Hill, New York). Tokura, Y., S. Koshihara, T. Arima, H. Takagi, S. Ishibashi, T. ldo and S. Uchida, 1990, Phys. Rev. B 41, 11657.

562

S.L. COOPER

Tomeno, l., M. Yoshida, K. Ikeda, K. Tai, K. Takamuku, N. Koshizuka, S. Tanaka, K. Oka and H. Unoki, 1991, Phys. Rev. B 43, 3009. Tranquada, J.M., D.J. Buttrey, V Sachan and J.E. Lorenzo, 1994, Phys. Rev. Lett. 73, 1003. Tranquada, J.M., J.D. Axe, N. Ichikawa, Y. Nakamura, S. Uchida and B. Nachumi, 1996, Phys. Rev. B 54, 7489. Tsuei, C.C., J.R. Kirtley, C.C. Chi, L.S. Yu-Jahnes, A. Gupta, T. Shaw, J.Z. Sun and M.B. Ketchen, 1994, Phys. Rev. Lett. 73, 593. Tsuneto, T., 1960, Phys. Rev. 118, 1029. Uchida, S., T. Ido, H. Takagi, T. Arima, Y. Tokura and S. Tajima, 1991, Phys. Rev. B 43, 7942. van Harlingen, D.J., 1995, Rev. Mod. Plays. 67, 515. Varma, C.M., EB. Littlewood, S. Schmitt-Rink, E. Abrahams and A.E. Ruckenstein, 1989, Phys. Rev. Lett. 63, 1996. Vekhter, I., J.R Carbotte and E.J. Nicol, 1999, Phys. Rev. B 59, 1417. Virosztek, A., and J. Ruvalds, 1991, Phys. Rev. Lett. 67, 1657. Virosztek, A., and J. Ruvalds, 1992, Plays. Rev. B 45, 347. Wake, D.R., E Slakcy, M.V. Klein, J.R Rice and D.M. Ginsberg, 1991, Phys. Rev. Lett. 67, 3728. Walstedt, R.E., WW. Warren, R.E Bell, R.J. Cava, G.E Espinosa, L.E Schneemeyer and J.V Waszczak, 1990, Phys. Rev. B 41, 9574. Warren, W.W., R.E. Walstedt, G.E Brermert, R.J. Cava, R. Tycko, R.F. Bell and G. Dabbagh, 1989, Phys. Rev. Lett. 62, 1193. Weber, WH., and G.W Ford, 1989, Phys. Rev. B 40, 6890. Wen, X.G., and EA. Lee, 1996, Phys. Rev. Lett. 76, 503. Wen, X.G., E Wilczek and A. Zee, 1989, Phys. Rev. B 39, 11413.

Wenger, E, and M. Kfill, 1997, Phys. Rev. B 55, 97. Wollman, D.A., D.J. van Harlingen, WC. Lee, D.M. Ginsberg and A.J. Leggett, 1993, Phys. Rev. Lett. 71, 2134. Wu, D.H., J. Mao, S.N. Mao, J.L. Peng, X.X. Xi, T. Venkatesan, R.L. Greene and S.M. Anlage, 1993, Phys. Rev. Lett. 70, 85. Wu, WC., and J.R Carbottc, 1997, Phys. Rev. B 56, 6327. Wu, WC., and J.R Carbotte, 1998, Phys. Rev. B 57, R5614. Yamamoto, K., T. Katsufuji, T. Tanabe and Y. Tokura, 1998, Phys. Rev. Lett. 80, 1493. Yamanaka, A., T. Kimura, E Minami, K. Inoue and S. Takekawa, 1988, Jpn. J. Appl. Phys. 27, L1902. Yoon, S., H.L. Lin, G. Schollerer, S.L. Cooper, P.D. Han, D.A. Payne, S.-W. Cheong and Z. Fisk, 1998, Phys. Rev. B 58, 2795. Yoon, S., M. Riibhausen, S.L. Cooper, K.H. Kim and S.-W Cheong, 2000, Phys. Rev. Lett. 85, 3297. Yoshida, M., S. Tajima, N. Koshizuka, S. Tanaka, S. Uchida and T. Itoh, 1992, Phys. Rev. B 46, 6505. Zawadowski, A., 1990, in: Electronic Properties of High To Superconductors and Related Compounds, eds H. Kuzmany, M. Mehring and J. Fink (Springer, Berlin) p. 306. Zawadowski, A., and M. Cardona, 1990, Phys. Rev. B 42, 10732. Zeyher, R., and G. Zwieknagel, 1990, Z. Phys. B 78, 175. Zha, Y., S.L. Cooper and D. Pines, 1996, Phys. Rev. B 53, 8253. Zhang, EC., and T.M. Rice, 1988, Phys. Rev. B 37, 3759. Zhou, X., M. Cardona, D. Colson and V Viallet, 1997, Physica C 282-287, 1007.

Handbook on the Physics and Chemistry of Rare Earths VoL 31 edited by K.A. Gschneidne~ Jr., L. Eyring and M.B. Maple © 2001 Elsevier Science B.V. All rights reserved

Chapter 204 CHARACTERIZATION OF CUPRATE S U P E R C O N D U C T O R S USING TUNNELING SPECTRA A N D SCANNING TUNNELING M I C R O S C O P Y H i r o h a r u S U G A W A R N ~'c*, Tetsuya H A S E G A W A b'c a n d K o i c h i K I T A Z A W N ~'~

a Department of Applied Chemistry and Department of Superconductivity, University o f Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan; b Materials and Structures Laboratory, Tokyo Institute of Technology, 4259 Nagatsutacho, Yokohama 226-8503, Japan; c Core Research for Exploratory Science and Technology (CREST), Japan Science and Technology Corporation (JST)

Contents

1. 2.

3.

4.

5.

564 Symbols 564 Acronyms 565 Introduction 566 Various junction structures 2.1. Artificial junctions for tunneling 566 measurements 568 2.2. Intrinsic Josephson junctions Electronic phase diagram and turlneling 570 process 570 3.1. Electronic phase diagram 3.1.1. Electronic phase diagram and 570 pseudogap 571 3.1.2. Spin gap and pseudogap 3.1.3. Energy and temperature scales of 573 the pseudogap 574 3.1.4. Anisotropy of the pseudogap 575 3.2. Tunneling process 576 Characterization of surface by STM 577 4.1. Bi2212 (001) surface 582 4.2. Y123 583 4.2.1. (001)-oriented plane 584 4.2.2. (110)-oriented plane 584 Tunneling spectroscopy 584 5.1. Superconducting gap structure

5.2. Temperature and doping dependence 5.2.1. Carrier concentration dependence of superconducting gap 5.2.2. Temperature dependence of superconducting and pseudogaps 5.3. Impurity effects 5.4. Inner vortex excitation 6. Phase-sensitive measurement 6.1. Dependence on surface orientations 6.1.1. In-plane anisotropy 6.1.2. Surface orientations and zero-bias anomaly 6.2. Josephson tunneling under magnetic field 6.2.1. Josephson tunneling across Y123/Pb junction in tfie ab-plane 6.2.2. Josephson tunneling across Y123/Pb junction along the c-axis 6.2.3. Splitting of zero-bias conductance peak in zero magnetic field 7. Summary Acknowledgement References

588 588 590 593 594 597 597 597 597 598 598 600 601 603 604 604

" Present address: Department of Electronic Systems Engineering, Tokyo Metropolitan Institute of Technology, 6-6 Asabigaoka, Hino, Tokyo 191-0065, Japan. 563

564

H. SUGAWARA et al.

Symbols El;

f(E) e

G(V)

I, J J J,

K kR N(O) N(E) H

P RN R~ T

T. L Tkw

Fermi energy distribution fimcfion electronic charge differential conductance at the bias voltage V [= (dl/dV)v] critical current tunneling current spin exchange energy current density critical current density wavevector of electron Boltzmann's constant density of states at the Fermi energy density of states at the energy E above the Fermi energy carrier density hole concentration per CuO2 unit momentum at Fermi energy electrical resistivity in normal state tunneling resistance temperature temperature below which the charge carriers pair up superconducting transition temperature temperature below which RVB state is obtained to result in anomalous metallic state

T~

T0'T',,

t t

VB g'samplc V'iP F A z~(/~)

App a(r) q~

~0 o(E) se Z

temperature below which a gap is thought to be open up in the spin excitation spectrum temperature below which various properties show anomaly in view of the Fermi liquid reduced temperature ( - T/T~) transfer energy in t-J model bias voltage sample voltage tip voltage lifetime of quasiparticle superconducting gap superconducting gap along the direction of k superconducting gap value defined at peak-to-peak superconducting gap at temperature T magnetic flux density of phonon states magnetic flux quanta tunneling conductance of quasiparticles at the energy E above the Fermi energy coherence length magnetic susceptibility

Acronyms AF AM ARPES BCS Bi2212 BTK DOS FL HTSC ITJ La214

anti ferromagnetic (phase or region) anomalous metal (region) angle resolved photoemission spectroscopy Bardee~Coope~Schrieffer Bi 2Sr2CaCu205+ Blondm~Tinkham-Klapwij k (formula) density of states Fermi liquid high temperature superconductor(s) intrinsic Josephson (tunneling) junctions La2_xSrxCuO4

NIN NMR PES RVB SC SG SIN SIS SQUID

normal metalqnsulator nnruaal metal (junction) nuclear magnetic resonance photoemission spectroscopy resonating valence band superconducting (region/phase) spin gap (region) superconductor-insulator normal metal (junction) superconductor-insulator-superconductor (junction) superconducting quantum interference device

TUNNELING AND STM CHARACTERIZATION OF HTSC CUPRATES STM

scanning tunneling microscope (microscopy)

UHV Y123

565

ultrahigh vacuurn YBa2Cu3Oy

STS

scanning tunneling spectroscopy

ZBCP

zero-bias conductance peak

T12223

T12Ba2Ca2Cu30~0 ~~

ZES

zero-energy states

1. Introduction

Tunneling spectroscopy is a powerful tool for exploring density of states near the Fermi level from measurements of the tunneling current versus the voltage between the sample and the counter metal electrode. Its energy resolution (< 1 meV) is higher than any other techniques, which is crucial for the experimental determination of the gap structures of superconductors. It has been quite successful for conventional superconductors: it proved the BCS mechanism for these materials and described the fine structures outside the gap edges on the basis of the electron-phonon interaction. This technique was expected to be useful also for the oxide high-critical-temperature superconductors (HTSC). A number of results have been reported so far, though their usefulness was rather limited compared to conventional superconductors because of the difficulty in obtaining an ideal tunneling junction on HTSC, which exhibit extremely short coherence lengths and inhomogeneous electronic structures. Tunneling results published in the early years after the discovery of HTSC were poorly reproducible, and there was little agreement as to which features were real in the superconducting gap spectra. Through efforts towards achieving better-characterized specimen and junctions, tunneling results have been converging, except for the details, in the last several years. This situation has been reviewed in several articles, e.g. Hasegawa et al. (1992a). At that time, the most important agreement was about the size of the superconducting gap. The superconducting gap values A are in the extremely strong coupling regime with reduced gap parameters 2A/kBT~ ~ 6-9 (optimal carrier density), compared to the BCS weak coupling value of 3.52. The BCS theory and its modified version were adopted to fit the spectra, though they failed to elucidate this large coupling strength. The observed tunneling spectra exhibited strange features. The differential conductance (dI/dV) inside the gap is finite, and sometimes even a peak appears (zero-bias anomaly). Various energy dependences, including linear and parabolic functions, were often observed for the background spectra outside the gap region. The origins of these features have been argued to be characteristic properties of these materials, though no clear answer was given. With the advent of cryogenic scanning tunneling microscopy (STM) it became possible to achieve vacuum tunneling gap conditions between sample and counter electrode, and to achieve atomic resolution on the surfaces of these materials. Vacuum tunneling is considered to be the best experimental situation, because it can avoid chemical reactions on the interface between the specimen and the insulating layer. In parallel to the above advancement, various physical properties of these materials have been intensively investigated, and many unique features of HTSC have been revealed.

566

H. SUGAWARAet al.

Above all, an order parameter other than s-wave symmetry has been argued for intensively in order to explain the experimental results from, e.g., NMR, penetration depth, specific heat, neutron scattering, and phase-sensitive SQUID measurements. Up to now, it seems that the argument has been converging to consider that the symmetry of the order parameter in these materials essentially has d-wave symmetry. Also, tunneling data have been discussed in the framework of unconventional pairing, and theoretical models have also been rewritten. It is claimed that several features such as the zero-bias anomaly observed in transverse geometry are indications of dx2 y2-wave symmetry of the order parameter (see sec. 6), while some of the latest investigations on YBaaCu3Oy show the admixture of s- and d-waves. It is also noted that an anomalous energy gap, now called the "pseudogap", opens up in HTSC above To. Tunneling spectroscopy has been one of the major tools for characterizing the pseudogap, which has added another mystery to the HTSC (see sec. 5.2). In this chapter, tunneling results published in the past several years are discussed with emphasis on STM results on Bi2Sr2CaCu2Oy (Bi2212) and YBa2Cu3Oy (Y123), the most intensively investigated HTSC compounds. The theoretical description and predictions closely related to the anisotropic superconductors are also examined. Following the comparison of various junction fabrication methods in sect. 2, characterizations of tunneling junctions by STM images are given. Superconducting spectra are then discussed from various viewpoints, and finally recent results on vortex imaging and phase-sensitive experiments are reviewed. 2. Various junction structures Tunneling spectroscopy is quite sensitive to the local electronic states formed at the surface. Therefore, the surface condition should be carefully examined. Unfortunately, the surface of oxide superconductors can readily be deteriorated in ambient atmosphere. It is pointed out that some of the Cu-based high-Tc superconductors lose oxygen easily, and an O-deficient surface region can be formed when the surface is exposed to an ambient low 02 partial pressure. It then loses its superconducting properties. The extent of the problem depends on the HTSC material itself. The problem is very severe in Y123, because of the large mobility of oxygen. On the other hand, Bi2212 compounds do not easily lose oxygen on the cleaved surface, although they are sensitive to electron-beam irradiation of the surface for incident-electron energies between 1 and 2 keV (Balzarotti et al. 1991). The surface nonstoicbiometry has made it very difficult to use inherently surfacesensitive analytical techniques like tunneling measurements, thus one should be always reminded that the surface electronic properties may be different from those of the bulk.

2.1. Artificial junctions Jbr tunneling measurements' The fabrication of the tunneling junction is an important experimental step in obtaining reliable measurements. The following four methods have been attempted for oxide superconductors:

TUNNELING AND STM CHARACTERIZATION OF HTSC CUPRATES

567

(1) Planar-type junction. An artificial insulating layer and a conducting layer (as a counter electrode) are deposited successively on the sample surface. (2) Point contact. A metal tip is pushed onto the sample surface. The native insulating layer either on the tip or the sample is used. The Schottky barrier formed at the interface can be used as well when a heavily doped semiconductor is used as the tip material. (3) Break junction. The sample is mechanically broken into two pieces, which are then brought close to each other so that a tunneling current is observed. In this geometry, both electrodes are the same superconductors, and hence Josephson tunneling is expected. (4) Scanning Tunneling Microscope. This is similar to the case of the point contact if the tip is in contact with the sample surface. However, the best performance is expected in the "vacuum" tunneling mode where the tunneling tip does not touch the sample surface. When the tip apex is composed of several atoms, tunneling takes place in a limited area on the surface, and a topographic image can be obtained by rastering the tip position on the sample surface. The planar-type junction and point-contact methods were first attempted on the HTSC. The reproducibility, however, was poor, and hence the results obtained by these methods need justification by the other methods. If a planar-type junction is ideally fabricated, it should be suited especially for performing measuring with high energy resolution and for measuring temperature dependence, because of the stability of the insulating layer. While the native oxide layer on conventional metal superconductors passivates the surface, this is not the case for HTSC. A short-circuit in a native or artificial insulating layer has been considered to be responsible for the appearance of zero-bias peaks accompanied by the proximity effect. Mechanical damage at the interface may also create small particles, causing a charging effect, which then gives rise to various spurious features in the spectrum such as multi-peak conductance. Another difficulty in pointcontact methods is the arbitrariness of the results due to the dependence on the contact pressure. Nevertheless, macrojunctions, e.g. between Pb and Y123, have been claimed to be successful, and are extensively used for some phase-sensitive measurements. It is also reported that good-quality planar junctions can be formed by atomic-layer epitaxial growth (Matsumoto et al. 1995). The break junction technique is also useful to acquire a clean interface, when the breakage is performed in situ in vacuum. Nevertheless, the resulting interface may be fractured into many pieces or into stepped structures, so one should be careful to check whether the momentum conservation is preserved or not. Contrary to the above techniques, a better junction quality seems to have been obtained with relative easiness by STM, because virtually no additional surface treatment is required. However, the application of this method is, in general, limited to easy-cleavage planes, or to natural as-grown surfaces.

568

H. SUGAWARA et al.

2.2. Intrinsic Josephson junctions The cuprate superconductors have layered structures in which, it has been pointed out, Josephson junctions are built within the layered structure itself. In Bi2212, each of the CuO2 bilayers stacking with a period of 1.5 nm acts as a pair of superconductor electrodes separated by an insulating layer. The formation of built-in Josephson junctions or intrinsic Josephson junctions (ITJ) along the c-axis in Bi2212 was claimed by Kleiner et al. (1992) and Kleiner and Mfiller (1994), who performed microwave emission experiments in small single crystals. Measurements of tunneling currents parallel to ITJ were then attempted for Bi2212 and TleBa2Ca2Cu3Ol0+ ~ (T12223). In earlier measurements using large and many (typically 1000) junctions, no gap structure was observed, possibly due to heating effects, in the current-voltage characteristics. A clear gap structure has been observed only with small and thin junctions in which a large tunneling current density can be achieved and heating effects can be eliminated (Kleiner et al. 1992, Kleiner and Mtiller 1994). The I - V characteristics of small stacked junctions of underdoped Bi2212 single crystals were investigated by Tanabe et al. (1996). The I - V characteristic of an intrinsic junction stack exhibits a superconducting branch and a large number of branches in the resistive state. This structure can be explained by assuming that the I - V characteristic of every individual junction exhibits a superconducting and a resistive branch, similar to the I - V characteristics of standard Josephson junctions. As shown in fig. 1, for a small bias current just above the critical current Io, the junction exhibits a periodic voltage jump. For a large bias current, larger hysteresis and negative resistance were observed. I

i

I

r

i

i

~

-I

0

1

1

i

t

.°.

4 -2 0

--

Contact Junction

Intrinsic Junction

I ~"

I', v"

V*

-2

75 -4 I - 0.4

i

~ - 0.2

i

I 0

I

0.2

r

I

0.4

V (V) Fig. 1. I - V characteristics o f a 300nm thick jtmction stack at 11 K. The inset shows the cutwe fbr a larger bias current. The critical current density Jc is 750 1250 A / m 2. The schematic cross-sectional view of the junction stacks is also shown. After Tanabe et al. (1996, fig. 1).

TUNNELINGAND STM CHARACTERIZATIONOF HTSC CUPRATES

569

The voltage jump Vj was found to be independent of the junction thickness and was 15-20mV. A quite similar value was reported by Kleiner et al. (1992) and Kleiner and Miiller (1994), and Schlenga et al. (1996, 1998) observed voltage jumps of about 22 mV in Bi2212 and about 23mV in T12Ba2Ca2Cu3010 ~ . The interval Vj of the step voltage was about 50% of the gap width (2z~) of Bi2212 observed by photoemission and other tunneling methods. Tanabe et al. (1996) attributed the effect of gap suppression primarily to nonequilibrium superconductivity. The value IoRN is a measure of the superconducting gap A because the relation IcRN = ~A/2e holds for a conventional energy gap. The IoRN product turns out to be about 5 mV, which is less than 20% of the maximum energy gap Ao/e. It is pointed out that the current-voltage characteristics below the gap are consistent with tunneling into d-wave superconductors. The tunneling quasiparticle current I(V) is calculated using the expression

I(V)

~

N(E)N(E-eV)(f(E-eV)

f(E)),

(1)

OO

where the normalized quasiparticle density of states N(E) for tunneling along the c-axis is given by

j

(2)

The I - V characteristic calculated based on a d-wave order parameter,

A( O) = Ao cos(20),

(3)

was in good agreement with the experimental results well below the gap, except for a deviation seen near the gap voltage. Inside the gap, dJ/dV is proportional to V2, which was claimed to be consistent with d-wave symmetry. Schlenga et al. (1996, 1998) reported the observation of subgap structures in I - V characteristics in step-edge junctions of T12Ba2Ca2Cu2010+~ and Bi2212. Well below the gap value --~50mV, the quasiparticle branch exhibits pronounced subgap structures, which were reported also by Yurgens et al. (1996). These structures were investigated as a function of several parameters such as temperature, direction and magnitude of applied magnetic field, and thickness of the junctions. Schlenga et al. claimed that the subgap structure was best explained by a resonant coupling mechanism between infrared-active optical c-axis phonons and an oscillating Josephson current proposed by Helm et al. (1997).

570

H. SUGAWARA et al.

3. Electronic phase diagram and tunneling process

3.1. Electronic phase diagram 3.1.1. Electronic phase diagram and pseudogap It has become increasingly important to understand the anomalous metallic state in order to understand the HTSC mechanism. Above all, the understanding of the pseudogap behavior in the hole-concentration vs. temperature phase diagram seems to be a key issue. The parent compounds of cuprate superconductors, which do not possess any holes, are antiferromagnetic insulators with a localized spin on each Cu site. Upon hole doping, which introduces holes mainly into the O 2p orbitals, the electronic system as a whole becomes mobile and metallic. The hole density p needs to be about 0.02 per CuO2 unit to destroy the long-range order of antiferromagnetism, and about 0.05 to induce the insulator-to-metal transition. Superconductivity also appears, starting at this composition, and T0 reaches a maximum a t p = 0.15, decreases again, and finally disappears a t p = 0.28, as shown in fig. 2. The same p-T~ dependence has been observed among all the cuprate superconductors. Many anomalous features deviating from Fermi-liquid behavior have been found in the region p < 0.28; these are called the "abnormal normal properties". The superconducting region in t h e p - T phase diagram is located in the low-temperature region of the "abnormal normal region", suggesting that high-temperature superconductivity manifests itself on this background. It has been experimentally revealed that the p - T phase diagram includes two additional pseudo-boundary lines, T* and To, as shown in fig. 2. The former was introduced initially as the line below which an energy gap seems to open up in the spin excitation, while the To line was defined as the temperature at which the static magnetic susceptibility Z exhibits a broad maximum, Xmax. The z - T curve was reported to be scaled well by )(max

"-x.

\

\ \

\to \

Anomalous Metal \

S p i n Gap ,..

0.05

0.15

0.28

Fig. 2. Commonly observed electronic phase diagram of the high-temperature superconductors. The superconducting (SC) region spreads over the hole concentration p per CuO2 unit from 0.05 to 0.28, showing the maximum Tc at 0.15. Below p - 0 . 0 2 , the antiferromagnetic (AF) phase is obtained. T o is the temperature below which various properties show anomalies with respect to the Fermi liquid, while below T* a gap is thought to open up in the spin excitation spectrum.

TUNNELING AND STM CHARACTERIZATIONOF HTSC CUPRATES

571

and To by Johnston (1989) and Nakano et al. (1994). Up to now the precise locations of the two lines are controversial, and besides it is not clear whether the two lines are independent or they just indicate a broad cross-over zone between them. In this context, the abnormal properties start being observed below To and more clearly below T*. 3.1.2. Spin gap andpseudogap The concept of the resonating valence bond (RVB) proposed by Anderson (1987) has stirred a great interest in the HTSC electronic structure. Anderson proposed that in the nearly half-filled band of strongly correlated electronic systems the electrons form spinsinglet pairs, leaving two types of quasiparticles; holons (charge) and spinons (spin). Suzumura et al. (1988) presented an electronic phase diagram, as shown in fig. 3, based on the t - J model, predicting T* as the spin gap line prior to the experimental findings. It was described that below Tc holons also pair up, creating Cooper pairs. Nagaosa and Lee (1990) introduced a gauge theory to relate the dynamics of holons and spinons and explained many of the "abnormal normal properties" observed experimentally by then, hence providing support to the RVB approach. It was Yasuoka et al. (1989) who suggested for the first time the presence of the "spin gap" based on experiment. They observed a sharp increase in the nuclear spin relaxation lifetime below T* (T* > T~) in the underdoped region. This was attributed to the opening of a "spin gap". They presumed that the spin freedom of electrons dies down below T*. Alloul et al. (1989) soon also assumed the spin gap from N M R Knight shift measurements. The existence of the spin gap was then more directly observed in neutron inelastic scattering spectra by Rossat-Mignod et aL (1991). As early as 1988, Ikuta et al. (1988) observed, with a point-contact junction, an energygap-like feature similar to the superconducting gap even up to room temperature in the tunneling spectra. It was Takagi et al. (1992) and then Nakamura and Uchida (1993) who noticed that the resistivity-temperature curve deviates from linearity and starts decreasing sharper towards lower temperatures below T*. More recent results were summarized by Nakano et al. (1994). On the other hand, Nishikawa et al. (1994) and Hwang et al. (1994) pointed out that the Hall coefficient starts deviating from typically metallic at To, as they tagged it, which was

V ~ / / / ~ TD ://A

0

AM Fig. 3. An electronic phase diagram predicted by the t-J model based on a mean-field slave boson approximation (Futa~yamadiagram). AF: antiferromagnetic region, AM: anomalous metal region, SG: spin gap region, SC: superconducting region, and FL: Fermi-liquid region. Below TIWBthe spin gap appears, while below TBthe charge carriers, i.e. holons, pair up. TB is the temperature below which a uniform RVB state is obtained, resulting in an anomalous metallic state.

572

H. SUGAWARA et al. i

. . . .

i

'

' ~ i

i,

,

.

i-'

'

'

'

i

. . . .

i

. . . .

L . . . .

i

, ~ " 3

t_.._a

-200

-100

0 Vsample

100

200

[mV]

Fig. 4. Tunneling spectra of the basal plane o f under-doped Bi2212 with Tc = 83 K at various temperataares. After Renner et al. (1998a, fig. 2).

much higher than T*, in correspondence with the temperature where the static magnetic susceptibility had been reported to show a broad maximum. Tallon et al. (1995) reported that thermoelectric power also showed the anomaly below To. From specific-heat measurements, Loram et al. (1993, 1994) observed an anomalous decrease in the electronic entropy at a much higher temperature than T* as well. According to them, however, the magnitude of entropy loss with decreasing T was larger than that expected from the formation of a spin gap, and hence some of the charge freedom had to be lost. Optical measurements are thought to sense charge excitation, but to be insensitive to the spin excitation. The presence of a charge gap was supported by Slakey et al. (1990), Homes et al. (1993) and Schlesinger (1994), while these observations have been disputed by Tajima et al. (1996) and by Basov et al. (1996) from their experimental studies. Recent STM tunneling results by Renner and Fischer (1995) and Renner et al. (1998a) have shown that the gap structure persists above ire and even near room temperature. Besides, the gap-like structure did not exhibit a large change across T~, but just grew continuously, as shown in fig. 4. Surprisingly, they also observed a nearly T-independent size, not only for the pseudogap in the normal region, but for the superconducting gap as well. They also succeeded in observing the vortex core for the first time by STM in the high-

TUNNELINGAND STM CHARACTERIZATIONOF HTSC CUPRATES

573

temperature superconductors (see sect. 5.4). in the case of Bi2212, they observed a gaplike structure in the vortex core which was similar to that of the pseudogap. Hence, they attributed the observed gap to the persistence of the pseudogap in the vortex-core region. As described above, there have been abundant results indicating the presence of some kind of energy gap which may be called either the pseudogap or the spin gap. Although the temperatures To and T* below which the anomalies are noticed are not well defined, they both exhibit a similar dependence upon the carrier density, i.e., going up as the carrier density decreases into the underdoped regime, as summarized by Nakano et al. (1994). The observation of the pseudogap has been mostly limited to the underdoped region, although some studies, especially performed by tunneling spectroscopy and specific heat on Bi2212, did report the extension of the pseudogap into the overdoped region. On the other hand, the To curve has been reported to be extending deep into the overdoped region not only in Bi2212 but also in Y123.

3.1.3. Energy and temperature scales of the pseudogap As results accumulated, it was a matter of concern how the characteristic temperatures T*, T0 and Tc, as well as the gap energy A, relate to each other, and how they depend on the carrier density. Although controversies still remain, it is interesting to note that Tc does not scale either with T* or T0. On the other hand, the latter two temperatures seem to rise in the underdoped region, while Tc falls. Another important indication is that the gap energy A remains rather constant as the temperature changes. Besides, A does not scale with T~, which it does for a BCS superconductor as 2A/kBT~ = 3.52. Oda et al. (1997a), who examined the details of the tunneling spectra and other transport and magnetic susceptibility results on Bi2212 (fig. 5), have pointed out that A rather closely scales with T* and To. They claimed that 2A/kBT* is 4 to 5, which is close to the predicted value for BCS d-wave pairing. According to them, T* and To are also related as To~T* being nearly equal to 2 for Bi2212. it is rather surprising that the gap energy in the superconducting state scales with T* or To, which are characteristic quantities in the normal state, but not with T~. However, experimental resuks are still contradicting each other in a quantitative sense, and hence reliability of the results must be examined critically. To start with, it is not experimentally clear whether T* and To are independent. Some interpret To as the initiation of short-range antiferromagnetic correlation, while others claim it to be the temperature where superconducting Cooper pairs are pre-formed. On the other hand, T* has been interpreted as the spin-gap temperature by some, while it is also claimed to be the temperature where long-range antiferromagnetic correlation grows but fluctuates. The region with pre-formed pairs is the "superconducting" region, but without phase coherence. Only after the development of phase coherence is the "zero-resistivity" superconducting state brought about. Hence, the abnormal metallic region is thought to be the phase-fluctuation region of superconductivity. The extremely large fluctuation is attributed to the low-dimensional structure.

574

H. SUGAWARA et al. Bi2212

Teo

(2Ao)

T*

STS(Thiswork) • • Raman (This work)

R-T (This work)

eli

300

(a)

200

La214 x7 Tunnel (Thb work) © CeI'T (This w°rk)

~ R-T (This work) O RH'T (Hwang et al.)

TN ~

~

T,&Tco

~'

T'aTco 100

-Is° 4 ...... •o.

Bi2212

.... °%o /"

0 0

J

0.1

I

P

0.2

/

" "~,

'l | O

0.3

Fig. 5. Dependence of Tc, T O (or Tin) and T* for La214 and for Bi2212, respectively. Data from Hwang et al. (1994) and Momono et al. (1999). After Momono et al. (1999, fig. 2a).

Another idea is that the antiferromagnetic correlation sets in at To and the system becomes abnormal. At a further, lower temperature T*, especially in the underdoped region, the long-range order develops. This may compete with or assist in the manifestation of superconductivity. Some even think that the antiferromagnetically ordered regions form stripes alternated with non-ordered conducting regions. The tunneling spectrum in any case is expected to exhibit a gap anomaly below To and more clearly below T*. It is hence highly important to further understand the nature of the pseudogap region in order to elucidate the superconducting mechanism. Tunneling measurements should be able to give the most precise information about the change in the gap feature with temperature and carrier density.

3.1.4. Anisotropy of the pseudogap Differing from the conventional s-wave superconductors, the symmetry of the Cooper pair in the HTSC has been established by now to be of d-wave symmetry. Its angular dependence was determined by ARPES (angle-resolved photoemission spectroscopy) to be dx2 y2 by Loeser et al. (1996) and (Ding et al. 1996). Furthermore, they have observed that the pseudogap above To also possesses dx~_y~ symmetry. These results

TUNNELINGAND STM CHARACTERIZATIONOF HTSC CUPRATES

575

make it tempting to assume that the mechanisms of creating the pseudogap and creating superconductivity are similar in origin. According to recent ARPES measurements, the pseudogap opens on the Fermi surface near k = (Jr, 0) or (0, Jr), called the "hot spots", while the surface near (½Jr, ½~) remains unchanged, "cold spots", in accordance with d~2_/ symmetry.

3.2. Tunneling process Let us consider electron tunneling through a metal (sample M1)-insulator (I)-metal (sample M2) interface, where the x and y-axes are defined parallel to the interface plane and the z-axis perpendicular to it. For simplicity, M2 is assumed to be an ideal metal with a spherical Fermi surface. If the interface planes are ideally flat, the wavenumbers of the tunneling electrons parallel to the plane, kx, ky, are conserved. In this case, referred to as "specular tunneling", the M I - I - M 2 tunneling mainly senses the excitation spectrum N( N EF, kz) of the sample with k~, because the tunneling decay length is proportional to ~zz. Thus, by fabricating tunnel junctions on different crystallographic specimens, the momentum dependence of N(E, fc) can be probed. If the interface has roughness of the order of the mean flee path, the tunneling electrons are scattered at the interface (diffusive tunneling), and the resultant tunneling spectrum is momentum-averaged, approximately equivalent to the density of states N(E). It seems that most of the tunneling processes through planar-type junctions on HTSC are close to the diffusive limit, so that anisotropy in the excitation spectrum can hardly be detected. In scanning tunneling spectroscopy (STS), on the other hand, the lateral (x,y) locations of tunneling electrons are established with an inaccuracy of the same order as the spatial resolution of STM, Ax, Ay < 1 nm. Thus, the obtained tunneling spectrum, in principle, suffers from averaging in the momentum space Xw(k)N(E,k), with an appropriate weight w(~:), which is determined by the shape of the tip apex. Once the specimen turns to a superconducting state, the obtained superconducto~ insulator-normal metal (SIN) spectrum probes the quasiparticle excitation in the superconductor, which directly reflects the symmetry of the order parameter A(k). If A(/~) has simple s-wave symmetry, as is realized in conventional low-temperature superconductors, one expects a finite gap of A with overshooting peaks just outside the gap in N(E), as illustrated in fig. 6. Even if A(~:) possesses anisotropic s-wave symmetry, a finite gap, corresponding to the minimum gap, appears. In dx2 y2 superconductors with A(/~) = cos 2kx - cos 2ky, in contrast, N(E) is gapless with linear N(E) for E << zl. It is noted that the extended-s wave A(/~) = cos 2kx + cos 2ky is also characterized to possess a gapless feature with two singularities at E = A1 and z~2. As has been pointed out by several groups, the tunneling matrix element M, which is the prefactor in the expression for the tunneling current, is also an important factor to determine the shape of gap structures in the STS tunneling spectrum of HTSC. Since HTSC possess highly directional covalent bonds, the tunneling probability is substantially dependent on the shape of the atomic orbital, to or from which the conduction electrons

576

H. SUGAWARA et al.

s-wave

dx2_y2-wave

Order Parameter

d/

' d--V

:

~" ~2A

2A i Normalized superconducting

. . . . i. . . . . I.

.

.

.

.

.

tunneling spectrum 0

~.V

0

V

Fig. 6. Differential conductance spectra expected for the tunneling from normal metal to superconductor with s-wave symmetry and d-wave symmetry.

tunnel. Accordingly, if the tunneling current dominantly flows along a specific direction, the tunneling spectrum is mainly contributed from N(E, k) for that direction (directional tunneling). Therefore, it is possible that STS tunneling spectroscopy sees a finite gap even in a dx2 y2-wave superconductor, if the tunneling electrons flow along the Cu O direction, in which Ax2 y2(/~) is maximized.

4. Characterization of surface by STM

Little consensus about features of tunneling spectra has emerged until recently, in spite of the fact that many tunneling spectra have been reported for Bi2212 (J.-X. Liu et al. 1991, Hasegawa et al. 1991b, Chang et al. 1992, Ichimura and Nomura 1993, J. Liu et al. 1994, Wolf et al. 1994, Rennet and Fischer 1995, Murakami and Aoki 1995, Oda et al. 1996, Mallet et al. 1996) and Y123 (Valles et al. 1991, Edwards et al. 1992, Miller et al. 1993, Nantoh et al. 1994, Maggio-Aprile et al. 1995). Various types of superconducting gap structures, as well as background (outside the gap) shapes, have been reported. Although the situation is quite complicated, it is reasonable to assume that the wide variety of results reported is due to the wide variety of physical and chemical conditions of the surfaces, insulating layer, and counter metal electrode, as well as bulk properties of specimen. In order to examine these results, one should carefully take the experimental conditions into account. On the other hand, STM is a powerful tool to investigate the tunneling spectra from this viewpoint, because a surface topographic image can be obtained during the measurement of the tunneling spectra.

TUNNELING AND STM CHARACTERIZATIONOF HTSC CUPRATES

577

In this section, STM images and tunneling spectra obtained by STM on Bi2212 and Y123 are reviewed. 4.1. Bi2212 (001) surface It has been pointed out that the surface o f Bi2212 is easily deteriorated in ambient atmosphere. In order to detect the intrinsic tunneling spectrum, the basal plane o f Bi2212, which is easily mechanically cleaved and relatively stable, has been investigated by numerous groups. The cleavage takes place between adjacent BiO bilayers, resulting in exposure o f the BiO plane on top, as demonstrated by room-temperature STM images reported so far (Kirk et al. 1988, Shih et al. 1988, Wang et al. 1990, Wu et al. 1990, 1991). In low-temperature STM images o f cleaved surfaces taken under a relatively large bias voltage t VBt ~> 0.1 V, one can clearly see the modulation structure along the b-direction with a period o f ~2.6 nm, as shown in fig. 7). The orthorhombic lattice with a0 ~ b0 = 0 . 5 4 n m is also observed, which indicates the topmost layer to be the B i - O layer. These images have also been reported by other researchers (Hasegawa et al. 1991 a,b, Manabe et al. 1994, Oda et al. 1996, Kaneko et al. 1998) Oda et al. (1996) reported STM images taken at lower temperatures under a much lower bias voltage VB = 20 mV As shown in fig. 8, a square atomic lattice with an interval o f ~0.4 nm and a weak modulation structure with a ~2.5 nm period along the diagonal direction were observed. They claimed that the S r - O and Ca planes are insulating and the tunneling current under low bias comes mainly from the metallic C ~ O plane, hence, the atomic corrugations observed in their STM image indicate a C u - O - C u network. Renner and Fischer's group reported a detailed low-temperature STM investigation (Barbiellini et al. 1994, Renner et al. 1994). On the surface cleaved under ultrahigh vacuum (UHV) conditions (~3 x 10 8 tort) at room temperature, STM measurements were made under helium gas atmosphere without exposure to air. (a)

5

(b)

o

5

?

o/~,

[nm]

5 o'.~3z b [nml

5

Fig. 7. Constant-current STM topographic images of in situ cleaved Bi2212 single crystals at 4.8 K (raw data). (a) Surface which shows a semiconducting gap (imaging parameter: It =275 pA, V~,,lpl~-0.5 V, Rt = 1.8 Gf2). (b) Surface which shows a semiconducting gap (It =300pA, Vsao3ple =0.5V~ i t = 1.7 G~). After Renner and Fischer (1995, fig. 1).

578

H. S U G A W A R A et al.

Fig. 8. Low-bias STM image of Bi2212 cleaved surface taken under constant height mode for V=20mV and initial current It = 1 nA at 6K. After Oda et al. (1996, fig. 2a).

aX,,/'b 400

. ' ' ' ' l

....

i ....

I ....

I ....

I ....

I ....

~ ' ' 1

....

I ....

I ....

I " ~

200 --30002 l '''l"''l~''l'''l'''l" ~ ~ r0

==

""300 ~

0

-200 (a) -400 80

l,,I,,,.I i'''l

....

....

i ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I,,,

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

:

' ° k ..................t _

<:

4O 0 -40

-300

0

/

300

(b) -80 -30(

I ....

I ....

-200

t ....

I ....

-100

I ....

I ....

0

I ....

I ....

100

Vs~pl ° [mV]

I ....

I ....

200

I.,,

300

Fig. 9. I - V characteristics of Bi2212 at 4.8K by Au tip, and differential conductance curves (inset) acquired by numerical differentiation. The numerical derivative of each I - V curve is shown in the inset. (a) Superconducting spectrmn acquired in the region of fig. 7a with Rt = 2.0 G~. (b) Semiconducting spectrum observed in the region of fig. 7b with R t = 1.0 G~. After Renner et al. (1994, fig. 2).

Topographic images with atomic resolution (fig. 7) were obtained at relatively higher t u n n e l i n g resistance (Vtip =0.5 V, I ~ 3 0 0 p A , R t ~ 1 . 8 G ~ ) . B y successive cleaving o f the same sample, they could o b t a i n either a superconducting or a s e m i c o n d u c t i n g spectrum (figs. 9a,b), while a n atomic-resolution i m a g e was taken. O n the other hand, a superconducting gap was obtained even w h e n atomic resolution was lost. Hence, they claimed that atomic resolution is neither necessary n o r sufficient to guarantee high-quality t u n n e l i n g spectroscopy. Surfaces cleaved u n d e r v a c u u m conditions were also e x a m i n e d b y K a n e k o et al. (1998). A c c o r d i n g to them, the t u n n e l i n g spectra can be grouped into three types, as s h o w n in

TUNNELING AND STM CHARACTERIZATION OF HTSC CUPRATES

°.°

579

-El

0.2

0

k)!,'

0.2 . . . . . . .

>

'

....

i''

0.1

0

°.°sV "(;) . . . . .

'"'"

°'°"r"~,_, -0.1

' ' ')t

,7"' 0

BIAS VOLTAGE

0.1 (V)

Fig. 10. Three types of differential conductance spectra typically observed on the cleaved Bi2212 surface. VB= 0.5 V and It - 0.1 nA at 4.2 K. After Kaneko et al. (1998, fig. 6).

"0

-200

0 200 V,amp~e[mV]

Fig. 11. Typical tunneling spectra observed on the as-cleaved surfaces of Bi2212 at 4.2K. Spectra are characterized by (a) a semiconducting gap, (b) a superconducting gap with flat background, (c) a superconducting gap with V-shaped background. The sample bias voltage and the set-point currents are: (a) 45mV and 2.51nA, (b) 49mV and 2.23 hA, (c) 56mV and 1.64hA, respectively. After Matsuura et al. (1998, fig. 2).

fig. 10: spectra exhibiting a clear gap structure w i t h a flat b a c k g r o u n d outside the gap; a gap with a V - s h a p e d b a c k g r o u n d ; and a s e m i c o n d u c t i n g or insulating gap structure. T h e s e three types o f spectra were o b s e r v e d irrespective o f w h e t h e r or not atomic i m a g e s were observable. M a t s u u r a et al. (1998) have recently found that the cleaved surface could be further r e m o v e d or " e t c h e d " by the S T M tip to create two types o f tunneling spectra, as shown in fig. 11, w h i c h is similar to those shown in fig. 10. M o s t o f the spectra r e v e a l e d a semi-

580

H. SUGAWARAet al.

Fig. 12. (a) STM image including an "etched" sm'face of Bi2212; (b) a cross-sectional profile of the a-b plane in (a). (c) Tunneling spectra observed on (A) the as-cleaved and (B) the etched surfaces. The sample bias voltage (V~ample)and the set-point current for (a) are 990 mV and 0.71 nA, respectively. The spectroscopic measurement positions are marked as crosses in (a). After Matsuura et al. (1998, fig. 3). conducting character with V-shaped or parabolic conductance. A clear superconducting gap was only occasionally observed, while the background outside the gap was found to be either V-shaped or flat with respect to the bias voltage, as shown in figs. 1 lb and 1 lc, respectively. Consecutive rastering on the sample leads to a removal or "etching" of such an insulating layer, as shown in fig. 12a, when the tip-to-sample distance is very small, e.g. in this case, the bias voltage is low, ~50 mV, and the set-point current is high, > 2 nA, corresponding to a tunneling resistance < 2 5 Mr2. In the etched region, a clear superconducting gap structure with overshooting peaks appears, as shown in fig. 12b. Removal or "etching" o f the surface was also reported in Bi2212 by Mallet et al. (1996), who observed at room temperature in a helium gas atmosphere that small clusters o f 1 0 0 - 3 0 0 A in size were cleaned by consecutive scanning on the sample. They speculate that these small clusters o f different oxygen content were formed by chemical reaction on the surface. A similar observation was reported by Harmer et al. (1991). Surface-dependent tunneling spectra have been observed by Murakami and Aoki (1995). They performed low-temperature STM/STS measurements on a cleaved surface o f Bi2212, and found three terraces eventually exposed by in situ cleavage. Here, as shown in fig. 13, the lowest and highest terraces had a one-step difference in height, and a square

TUNNELING AND STM CHARACTERIZATIONOF HTSC CUPRATES

581

Fig. 13. (a) STM image (image size: 56nmx56nm) of a Bi2212 cleaved surface; (b) a cross-sectionalprofile along the line A and J in the topograph (a). After Murakami and Aoki (1995, figs. la,c).

lattice arrangement was observed in the STM image, indicating the exposure of BiO layers. The intermediate terrace, which did not give an atomic image, hence, was unlikely to be a BiO layer. They observed quite different ttmneling behavior on the top/bottom planes and intermediate terrace, as shown in fig. 14. On the highest and lowest terraces, although a clear superconducting gap was observed, the spectra taken under higher bias were semiconducting. On the other hand, in the intermediate-terrace region, the superconducting spectra showed a gap with rather a flat bottom. They speculated that in the intermediate terrace, a Ca-layer site between the two C u Q layers was likely to be exposed. The results described above suggest that tunneling spectra from cleaved surfaces still differ from one to another. Nevertheless, several researchers (Renner and Fischer 1995, Kaneko et al. 1998) have claimed that reproducible spectra are obtained when the cleavage and the measurements are performed under very inert conditions. They agree on some important features: vanishing differential conductance at zero bias, and flat background. Such reproducible spectra do not change their shape regardless of the measurement position and the tip-to-sample distances. It has been questioned whether tunneling spectra are dependent on the tip-to-sample distance, a quantity experimentally determined by bias voltage and set-point current.

582

H. SUGAWARAet al.

,-7.

,d >

-IOO

o

V[mV]

Ioo

Fig. 14. STS spectrum characteristics at different points in the mid-terrace region of the cleaved surface of the same Bi2212 sample used in fig. 13. After Murakami and Aoki (1995, fig. 6). Hasegawa et al. (199la), Ichimura and Nomura (1993), and Murakami and Aoki (1995) reported a strong dependence on the tip-to-sample distance, while the others did not. V-shaped background spectra have been reported by many research groups (Renner and Fischer 1995, Honcotte et al. 1995, Mallet et al. 1996, Kaneko et al. 1998), and their results have been discussed in terms o f an anomalous normal-state property o f the material (Anderson and Zou 1988, Flensberg et al. 1988, Varma et al. 1989, Kirtley and Scalapino 1990). 4.2. 11123 Observation o f an atomic image in Y123 was less frequently reported than in Bi2212. This seems to be due to the fact that there is no stable cleavage surface in a bulk single crystal. In addition, the fractured surface is chemically unstable in air, and oxygen is released under low-pressure oxygen conditions. According to Edwards et al. (1994b), oxygen release is suppressed when the cleaving is performed below 60 K. Nevertheless,

TUNNELING AND STM CHARACTERIZATIONOF HTSC CUPRATES

583

the atomic arrangement and the superconducting gap structure has been observed on the surface o f single crystals or thin films. 4.2.1. (O01)-orientedplane On the surface o f epitaxial Y 123 thin films, several topographic features o f crystal growth have been observed, such as spiral (Hawley et al. 1991, Lang et al. 1991), ledge (Moreland et al. 1991), and needle (Muenchausen et al. 1992) types, which indicates that the surface during the growth o f thin films is much more chemically stable (Lang et al. 1991) than the fractured surface (Narlikar et al. 1992, 1996). Hasegawa et al. (1993) and Nantoh et al. (1994) have observed Y123 (001)-oriented thin epitaxial films grown by laser ablation. As shown in fig. 15, the STM image taken at 4.2 K clearly shows atomic arrangement with an average spacing o f 0.35 nm.

Fig. 15. STM hnage of a (001)-oriented Y123 thin film taken at 4.2 K. (a) Vui,~= 500 mM It - 50 pA, imaging size: 7x7nm 2. (b) Vbi,~=500mV, I~=15pA, imaging size: 4.5x4.5 nm2. A square surface lattice with average spacing of 0.4nm corresponds to the a- or b-spacing of the unit cell. After Nantoh et al. (1994, figs. ld,e).

584

H. SUGAWARAet al.

According to Edwards et al. (1992, 1994a,b, 1995), a clear atomic image could be obtained for Y123 single crystals grown by the flux method only when the cleavage and the measurement were performed below 60 K. Their images show one-dimensional structures, which they claimed to correspond to the CuO chain. The one-dimensional structure had a limited length up to about 13 A, and appeared to form dimers. The electronic character o f the surface was semiconducting, and hence they claimed that a finite gap appeared in the tunneling spectrum due to the formation o f a charge-density wave. Maggio-Aprile et al. (1995) observed no atomic image on single crystals, while a superconducting spectrum could be detected on the surface. They succeeded in observing variation o f the tunneling spectrum in the vortex-core regions (see sect. 5.4). 4.2.2. (110)-oriented plane it is believed that a large anisotropy should exist along direction o f the C u - O chain, but Nantoh et al. (1995) reported a controversial result. Figure 16 presents STM images o f (110) surfaces o f Y123 epitaxial films, in which an orthorhombic lattice of surface atoms is clearly observed. The image of the CuO chain is brighter than that o f the CuO2 layer, and the contrast is enhanced for lower bias voltage. This implies that the density of states at the Fermi level is higher in the CuO chain than in the CuO2 layer. The semiconductive nature of the CuO chain is claimed to have smaller anisotropy as compared to other HTSC materials.

Fig. 16. STM images of a (ll0)-oriented Y123 epitaxial film taken at 4.2K. (a) Vtiv--1.9V, /I 0.13hA. (b) Vtip=-0.75 V, I~= 0.09 hA. Imaging size: 6.5 x 6.5 nm2. After Kawasaki and Nantoh (1994).

5. Tunneling spectroscopy 5.1. Superconducting gap structure The details o f the superconducting gap structure form a rich source of information about the pairing mechanism. If the pairing is o f typical s-wave symmetry, a finite gap is

TUNNELING AND STM CHARACTERIZATION OF HTSC CUPRATES

585

opened and hence the in-gap conductance is very low. If it is of p- or d-wave type, the gap is not completely opened and hence the conductance is finite, decreasing with a certain slope towards zero bias. In conventional superconductors, finite gaps have been observed in support of the s-wave. In the case that the electron-phonon interaction is sufficiently strong, a fine structure is observed in the tunneling spectrum outside the gap. Detailed analysis of the fine structure has provided the spectrum of the electron-phonon interaction which gives the most decisive proof for the phonon-mediated superconducting mechanism by comparing it with the independent phonon spectrum determined by neutron scattering. In the HTSC, although some fine structure has frequently been observed, there has been little understanding of the fine structure, and experimental consistency has not been fully examined. So far the greatest attention has been paid to the main gap structure: gap size and in-gap conductance. In the initial period of studies on HTSC, tunneling spectrum measurements were quite inconsistent from one to another, as summarized by Hasegawa et al. (1992a). Multiplepeak and multiple-gap structures were frequently reported feeding speculations about the exotic nature of the HTSC mechanism. However, especially owing to the development of the STM tunneling method, the results gradually converged to the rather simple single-gap structure. This was because the junction under the STM tip was thought to be undamaged compared to the other techniques. Therefore, experiments with other methods started utilizing the STM results as a standard to judge the quality of their junctions. STM techniques in the early days were, however, immature with respect to the limitations of atmosphere, temperature and stability against noise. The preparation of the surface was another problem. By now, several groups have achieved in situ cleavage of the sample, or at least cleavage in high vacuum. Also, it has been found that in situ etching under certain bias conditions provides a surface on which reproducible measurements can be performed. These results have been reviewed by Kitazawa et al. (1996) and by Sugawara et al. (1996). The controversial issue then was whether the gap feature was s-like or d-like. Comparing the available tunneling spectra on Bi2212 and Y123, Kitazawa et al. (1996) noted that there were essentially two kinds of results: one exhibiting a clear gap, as in a BCS metallic superconductor with a rather flat bottom, the other exhibiting a gap with a large zero-bias conductance and/or a sharp V-shaped conductance close to zero bias, as shown in figs. 17 and 18. The former type was reported by Hasegawa et al. (1991b), Hasegawa et al. (1992b) and by Murakami and Aoki (1995); they observed this type of gap only on a non-BiO surface produced accidentally on a part of the sample. On the other hand, the latter type was reported as well by these authors, and by many other groups: Renner et al. (1994), Manabe et al. (1994), Honcotte et al. (1995), Maggio-Aprile et al. (1995), Ueda et al. (1995), Mandrus et al. (1993) and Matsuura et al. (1998). Hence, the latter type was the typical commonly observed spectrum for the HTSC, although some experimental ambiguities remain as to why the spectra appeared differently on different surfaces.

586

H. SUGAWARA ct al.

-100

-50

0

50

100

V (mV)

Fig. 17. Tunneling spectra on the basal plane of Bi2212 and Y123 with the terminating layer claimed to be a CuO2 plane, or a SrO (Bi2212) or BaO (Y123) plane covering the CuO2 layer. Curves H-N-K (Bi) Hasegawa et al. (1991b), H-N-K (Y) Hasegawa et al. (1991b, 1993), Nantoh et al. (1991, 1994, 1995), M A (Bi) Murakami and Aoki (1995) are for Bi2212, Y123, and Bi2212, respectively. After Kitazawa et al. (1996, fig. 1).

H-N-K ,.--7,. :El

_

- 100

I

R - F (Bi) I

I

I

-50

0

50

v (mv)

00

Fig. 18. Tunneling spectra on the basal plane of Bi2212 and Y123 with the terminating layer claimed to be a BiO and CuO chain plane, respectively. The references for the curves are: H-N-K, Hasegawa et al. (1991b, 1993), Nantoh et al. (1991, 1994, 1995); R-F, Renner and Fischer (1995); M O I, Manabe et al. (1994, 1997); and M-A (Bi), Murakami and Aoki (1995). After Kitazawa et al. (1996, fig. 2).

It was difficult to interpret the latter type o f spectrum in an s-wave framework, unless a rare case such as extended s-wave was assumed. In the m e a n t i m e , m a n y other observations m a d e by different methods such as N M R , specific heat, phase-sensitive S Q U I D method, A R P E S , etc., supported d-wave symmetry, a n d as a result, t u n n e l i n g spectra o f this k i n d were analyzed according to d-wave s y m m e t r y and are n o w regarded m o r e positively as one category o f evidence to support it. SIS-jtmction spectra observed on the naturally occurring built-in layer structure o f Bi2212 single crystals have b e e n reported by several groups. These spectra also give rather high and linearly increasing in-gap conductance, again in accord with the d-wave pairing m e c h a n i s m (Tanabe et al. 1996, Schlenga et al. 1996, 1998). Kashiwaya and Tanaka (Kashiwaya et al. 1994b, 1995b, 1996, Y. T a n a k a and Kashiwaya 1995) have e x a m i n e d both theoretically and experimentally the interface effect o f anisotropic superconductors, predicting the occurrence o f the zero-bias c o n d u c t a n c e peak and its splitting in some cases due to the formation o f A n d r e e v b o u n d states

TUNNELINGAND STM CHARACTERIZATIONOF HTSC CUPRATES

587

at interfaces (or surfaces). Their systematic results were again consistent with dx2 p2 sytmnetry (see sect. 6). Therefore, as far as the symmetry of the Cooper pair is concerned, it is thought that d-wave symmetry has been well established. The differences between experimental results then seem to be attributable to surface- and/or angle-dependent phenomena. The second interesting issue is the size of the gap, especially the gap dependence on temperature and on composition. Ikuta et al. (1988) was the first to report that a superconducting-gap-like feature persists in the HTSC well above the critical temperature. Detailed results by Renner et al. (1998a) and Matsuda et al. (1999) have indicated that the change in the gap size in the superconducting temperature range is small. In addition, they note that the gap survives continuously in the tunneling spectrum above Tc. Interestingly, the gap gradually vanishes in that the gap is gradually filled up rather than closed, even above To. This temperature dependence is quite surprising, considering that in a BCS superconductor the gap decreases sharply as the temperature approaches To, and closes there. This is in accord with the idea that the bonding strength of the Cooper pair is also lost at To. In the HTSC, however, the observed temperature dependence rather suggests that the Cooper pair keeps the same binding strength even at Tc while the total number of pairs decreases gradually. However the pairs do not totally disappear at To, surviving to much higher temperature. It is quite important to know whether pairs above Tc can ever be Cooper pairs. At the present moment, we do not have a clear answer to this important question. However, it should be quite difficult to assume completely different nature for bound pairs below and above To. If the pairs below and above T~ are both Cooper pairs, then Tc should not be determined by the binding force, as is the case for conventional superconductors. Emery and Kivelson (1995) have proposed the idea of "pre-formed pairs", with their hypothesis that Tc is the temperature where complete coherence is established among the Cooper pairs. T~ decreases towards underdoping, because the density of Cooper pairs decreases and establishment of coherence is more difficult. The model described above is challenged mainly by two other ideas. One is to assume that the pair above T¢ is a "spin singlet" in the framework of spin-charge separation of an electron. Tc then is the temperature where the charge part of the freedom of an electron also pairs up, i.e. holon pairing. Another model assumes the pair formation to be electron pairing on the Fermi surface, but taking place in different ways on the surface: near/~ = (Jr, 0) at much higher temperatures, but near (½Jr, ½~) at much lower temperature. Only when sufficient numbers of electrons form pairs does superconductivity set in according to the latter model. At any rate, it is then clear that the scaling rule 2A/T~ ,~ 3.52 should not be observed in HTSC. Indeed it has been reported that T~ decreases towards underdoping, while the gap size remains nearly the same (Renner et al. 1998a) or even increases (Miyakawa et al. 1998). This will be discussed further in section 5.2. As for Y123 crystals, no consensus has emerged about the intrinsic tunneling spectra. Maggio-Aprile et al. (1995) investigated tunneling spectra on a fully oxygenated Y123

588

It. SUGAWARAet al. 2.0

I

. . . .

I

'

'

'-'r'-l-~'

'

I

. . . .

I

. . . .

I

. . . .

I

. . . .

I

. . . .

1.5

© ;> "c3

1.0

0.5 . . . . . . . . . ' . . . . ' . . . . . . . . . . . . . . . . . . . . . . . . -100 -50 0 50 Vsample [1T1V]

100

Fig. 19. Differentialconductance spectrmnof an Y123 single crystal at 4.2 K. After Maggio-Aprileet al. (1995, fig. 1). twinned as-grown single crystal from growth in a BaZrO3 crucible, which did not react with the reactive melts used as solvents for Y123. As shown in fig. 19, the tunneling spectrum taken at 4.2K showed gap-like features, with main conductance peaks at 20-4-2 inV. The background conductance outside the gap increased linearly with increasing bias. Note that the zero-bias conductance amounted to about 60-70% of the normal conductance, which was reproducible for various tip-to-sample distances and various vertical measuring positions. Similar features were also observed with a planar junction by Valles et al. (1991). The tunneling spectra depicted in fig. 19 are quite different from those in Bi2212 single crystals. The gap structure is not consistent with an isotropic BCS s-wave state, and is different from the simple d-wave picture. The multi-peak structure resembles the spectrum expected for extended s-wave, except for the large zero-bias conductivity. 5.2. Temperature and doping dependence 5.2.1. Carrier concentration dependence of superconducting gap In BCS weak-coupling superconductors, the magnitude of the superconducting gap A

is proportional to T~, following 2A = 3.52kBT~. The correlation between A and Tc in HTSC was first investigated by Tsai et al. (1989). They performed point-contact tunneling spectroscopy on Y123 thin films with various To, which were prepared by controlling the film growth conditions. As a result, they confirmed a linear relationship A ec Tc. Recently, however, several groups have attempted systematic tunneling measurements on a series of Bi2212 specimens with different carrier concentrations, n, reporting the possibility that A does not scale with Tc at all, but rather is governed by other temperature scales, such as T* and To. Momono et al. (1999) performed STS tunneling spectroscopy on La214 and Bi2212 with different doping levels. As shown in fig. 5, they found that for both systems

TUNNELINGAND STM CHARACTERIZATIONOF HTSC CUPRATES 1 ....

t ....

I ....

I ....

I ....

l ' ~ " r

....

l

T =56.0K 0 j l

....

h . . . W

....

i ....

589

,~

t

v_2_

-300 "~

-150

0

150

300

Vsample [1TIV]

Fig. 20. Tunneling spectra of Bi2212 at 4.2K for different oxygen doping levels. After Renner et al. (1998a, fig. 1). the zl values increased monotonously with decreasing carrier concentration n, through underdoped to overdoped regions, which is consistent with the results of Raman measurements by Kendziora and Rosenberg (1994). That is, A in the underdoped specimen is larger than that of the optimally doped one with the highest Tc. Interestingly, the obtained A value scales as To, as determined from magnetic measurements, although the scaling factor, r = kBTO/A, differs for the two HTSC systems: r ~ 5 for Bi2212 and r ~ 1 for La214. They also found that the reduced gap 2A/kBT* for different doping levels fell in a narrow range of 4-5, which is comparable to the mean-field value for the d-wave pairing state, ~4.3, leading to the conclusion that the temperature T* gives the mean-field values of To. This situation favors the pre-formed pair model that the origin of the anomalous behavior around Tc is essentially identical to that of superconductivity. Experimentally, the correlation between A and n is controversial at the present stage. Renner et al. (1998a) reported STS tunneling spectroscopy on Bi2212, where the specimens were cleaved in UHV. Figure 20 qualitatively illustrates the relationship between n (the oxygen content) and zl at 4.2K. it is notable that A increases as n is lowered, similar to the STS results by Oda et al. (1997b). Also see the comparison of tunneling, photoemission, and penetration depth data given by Panagopoulos and Xiang (1998). Based on quite stable STS measurements on Bi2212, however, Matsuda et al. (1999) advocated a different A-n correlation. They reported A as a function of oxygen content x, which is considered to be a measure of n. As shown in fig. 21, A is maximized around optimal x ~ 8.25, where Tc is also maximized, although A does not show a strong

H. SUGAWARAet al.

590 70

130 [3

120

6O

110

5O 40 "~

30

90 80

~

20

70 J 12 ~ ~2 L I ~ 10 8.21 8. 3 8. 5 8.27 8 . 2 9 OxygenContent

Fig. 21. Oxygen content dependence of the critical temperature (T~)and the superconductinggap value A. After Matsuda et al. (1999, fig. 1).

x-dependence in the underdoped region, in these observations, anyway, there exists no correlation between A and Tc. Miyakawa et al. (1998) reported tunneling measurements using a break junction on Bi2212 with various oxygen contents. They found a trend that A decreases monotonously with increasing n over a narrow doping region where Tc exhibited a maximum. The experimental inconsistencies described above may be due to the inhomogeneity of n, especially for underdoped samples. Further detailed studies are desired. 5.2.2. Temperaturedependence of superconducting and pseudogaps In the mean-field theory of superconductivity, the order parameter A is almost constant up to the reduced temperature t = T/Tc ~ 0.5, and then goes to zero as T approaches To. For HTSC, the temperature dependence of A reported so far can be classified into two groups. In the first category, the peak-to-peak separation energy 2App follows the BCS prediction, while in the second category 2App increases as the temperature is raised. Here, it should be emphasized that Avv does not correspond to A at higher temperatures, because the thermal smearing effect tends to shift App to the higher-energy side. In the superconducting states, the energy gap A(T) opens up at the Fermi energy, so that the quasi-particle density of states, N(E), for E > A(T) is higher than that for E > A(T). Thus, a smearing phenomenon, in general, including lifetime broadening as well as thermal broadening, causes the upper deviation of App(T) in the A(T)-T curve. The extent of deviation depends on the magnitude of A and the shape of the N(E) function inside A. In conventional weak-coupling BCS superconductors, indeed, App(T) increases monotonously with T. The experimental disagreement for A(T), described above, may be explained by differences in the apparent tunneling conductance a(E) inside the gap. For example, if Andreev reflection takes place, (7(E) for E < A is apparently enhanced, compared with N(E), and thus, Z~pp is less influenced by the smearing effects. Recently, STS measurements on clean surfaces of Bi2212 prepared in situ have shown a slight increase of A at higher temperature, as plotted in figs. 4 and 22. Matsuda et al. (1999) analyzed the observed tunneling conductance curve, assuming a d-wave gap function and lifetime and thermal broadening effects, and demonstrated that A increased

TUNNELING AND STM CHARACTERIZATION OF HTSC CUPRATES

591

4

---r/l

3

r.,¢l

0

2

Z

1

0 -300

-200

-100

0

100

200

300

Sample V (mV) Fig. 22. Temperature dependence of tunneling DOS for the optimum doped Bi2212 smnple (T~ ~ 90K). Data are normalized to 2.5 at - 2 5 0 m e V After Matsuda et al. (1999, fig. 4).

with temperature, as shown in fig. 23. They also claim that the damping parameter F is rapidly enhanced near Tc, which is qualitatively consistent with the results of microwave response and optical measurements. In conventional superconductors, as mentioned above, A closes at T = ~, and the sign of A appears only as the superconducting fluctuation for T > Tc. On the contrary, it is a quite distinctive and unique feature for HTSC that the gap structure still remains up to temperatures significantly above To. In underdoped Bi2212, Renner et al. (1998a) observed a temperature dependence of tunneling spectrum over a wide temperature range between 4.2 K and room temperature. They found that a pseudogap, equivalent to A at 4.2 K, still remains even above Tc ~ 83 K, and does not change much with temperature up to room temperature. This observation is in striking contrast to the PES results, which indicate the reduction of the pseudogap around T* = 170K for the underdoped sample. Notably, Renner et al. (1998a) have reported a similar pseudogap structure even for the overdoped sample at least up to 80K above Tc =74.3 K. This also disagrees with the conclusion obtained from PES that the pseudogap for T > T~ can be seen only in the underdoped region. Recently, Matsuda et al. (1999) have performed a more detailed study on the pseudogap of Bi2212 as a function of temperature and carrier concentration. All specimens

592

H. SUGAWARA et al.

100

. . . .

un

i

. . . .

t

. . . .

i

. . . .

~

. . . .

t

. . . .

Result of BCS fitting 8.22. . r r l

80 > --- 60

o,t m um "

<1

40 8.28 overdoped super-gap pseudogap 20

. . . .

0

I , ,

I

. . . .

100

I

. . . .

I

. . . .

200

Temperature (K)

I

. . . .

I

300

Fig. 23. The fitted results for A for several doping levels. After Matsuda et al. (1999, fig. 5a).

investigated showed a pseudogap structure even at room temperature, although it tends to be smeared out more rapidly with temperature in the overdoped samples. The inconsistency between STS and PES can be interpreted in the following ways. Firstly, both probes, which possess different probing energies and are based on different excitation mechanisms, may sense different aspects of the excitation spectrum of HTSC. Secondly, PES has poorer energy resolution ALE~ 20 mV in comparison with STS. Thus, PES may be less sensitive to the change of pseudogap structure, particularly in the case that N(0) is continuously increased without closing the gap at higher temperatures. Thirdly, the origin of the inconsistency may be the slight loss of oxygen atoms from surfaces, resulting in decreased n in the surface layer. STS, being a more surface-sensitive technique, is expected to be more affected by the local reduction of n. It should be stressed here that at a first sight the gap structure shows no discontinuous change at Tc. It seems to vary continuously as a function of temperature. On careful inspection of figs. 4 and 22, one may recognize a slight discontinuity across T~. Figure 24 is a 3D plot of fig. 4, in which the energy value Ep giving maximum tunneling conductance is traced as function of temperature. As can be seen, E/, jumps abruptly at To. The result of curve-fitting analysis by Matsuda et al. (1999) more directly indicates that the A value is discontinuously enhanced at Tc with rising temperature (fig. 23). Consequently, they claimed that the pseudogap above Tc and the superconducting gap below Tc come from different origins, and compete with each other. In this scenario, then, the latter overcomes the former at T~. However, the discontinuity in Ep or A at Tc may result from a substantial increase of quasiparticle damping just below Tc, which smears the tunneling spectrum. As mentioned earlier, the smearing effect tends to shift Ep to the higher-energy side, and could cause a larger uncertainty in A estimated from the curve-fitting procedure.

TUNNELINGAND STM CHARACTERIZATIONOF HTSC CUPRATES

593

Fig. 24. (a) Three-dimensionalview of the conductance data shown in fig. 4. (b) Projection onto the energy temperature plane. The line at positive bias indicates the position of the positive-biasconductance peak which clearly shifts to higher energies above To. After Renner et al. (1998a, fig. 3). 5.3.

Impurity effects"

The variation of the superconducting gap with respect to T0 is an issue to be investigated in view of the understanding of the mechanism of high-Tc superconductivity. It was found in earlier experiments that the influence of impurity atoms on Tc in HTSC is different from that of conventional superconductors. For example, the substitution of Zn, a nonmagnetic ion, in La2 xSrxCuO4 (La214) compounds leads to the decrease of Tc to nearly the same extent as that of magnetic Ni. Nonmagnetic impurities induce little suppression of superconductivity in conventional superconductivity (Anderson's theorem of disorder). On the other hand, magnetic impurities like Ni and Fe in Y123 compounds yield a smaller Tc suppression compared to Zn. Numerous experiments indicate that effective substitution at the Cu site in the CuO2 plane leads to significant changes and to a reduction of To. However, only a few studies on tunneling spectroscopy are available. Broekholt et al. (1992) reported tunneling measurements by the point-contact method for Bi2212 with the substitution of Co for Cu. With increasing Co content, the transition temperature To decreased monotonically, while the superconducting gap d decreased more rapidly than To; the reduction at higher Co content was much stronger than the relation 2A/kBTc ~ 5.7 which they found for oxygen-deficient pure Bi2212 single crystals. Hancotte et al. (1997) investigated Zn- and Ni-substituted Bi2212 single crystals by STM and break-junction methods. While their pure sample showed T~=85K,

594

H. SUGAWARAet al.

Ni-substituted (content 1.3-1.5%) sample showed T~= 75 K. The superconducting gap value varied more significantly than for the pure sample, and the prominent gap value was 2Z~pp ~-, 77 meV, which was larger than the pure-sample value 2App ~ 52 meV, indicating increased coupling strength. The temperature-variant measurement by means of the breakjunction method revealed that the 2App value did not decrease monotonically with respect to the measuring temperature T but slightly increased in the reduced temperature range T/Tc = 0.3-0.5. On the other hand, the substitution of Zn (content ~1%) showed Tc ~ 79 K, and the gap value 2App ~ 50 meV was almost the same value as that of the pure sample. However, the conductance at zero bias was increased. This increase, estimated to a residual DOS of ~0.4, was in good agreement with the results of NMR experiments (Ishida et al. 1993). Their results were compared with theoretical results of impurity effects in d-wave superconductors in the unitary limit (Balatsky and Salkola 1996, Salkola et al. 1996), though they did not claim that the zero-bias anomaly observed in Zn-substituted Bi2212 is explained by their prediction. These variations in the substituent atoms seem to indicate that Co, Zn and Ni affect the electronic states in different ways. Note that Kane et al. (1998) recently investigated Pb-doped Bi2212 single crystals, and indicated that the in-plane energy gap was little influence by lead doping up to 4%, in spite of decreasing Tc. 5.4. Inner vortex excitation Another issue to be investigated is the excitation spectrum inside the vortex core, where the penetration of the magnetic field leads to local breakage of the time-reversal symmetry, resulting in local reduction of the order parameter. STM is a suitable tool to detect the spatial variation of the excitation spectrum inside the core. In the past, this technique was applied to the investigation of vortex lines in 2H-NbSe2 by Hess et al. (1989) and 2H-Nbl xTaxSe2 by Renner et al. (1991). At the vortex core, the differential conductance (d//dV) at zero bias voltage showed a peak, which gradually disappeared when the measurement position moved away from the core center, while small peak structures appear at finite bias voltages. Finally, far from the vortex-core center, the excitation spectrum becomes coincident with the bulk spectrum. It was found that the peak structures correspond to bound states of quasiparticles confined in the vortex core, and they were reproduced by a calculation based on the quasiclassical Eilenberger theory of the BCS s-wave order parameter (Caroli et al. 1964). Furthermore, mapping of the ratio of the conductance values at different bias voltages revealed a vortex lattice structure. In NbSe2, the vortex-core structure had 6-fold symmetry, which was explained by the effect of the crystal field. It was pointed out that the symmetry of the order parameter, either s-wave or d-wave, should be clarified from tunneling conductance in the vicinity of zero bias. The excitation spectrum of an s-wave superconductor forms discrete bound states, which are equally spaced to A/EF. On the other hand, several authors (Won and Maki 1996, Maki et al. 1996, Schopohl and Maki 1995, Ichioka et al. 1996) predict, by solving the Eilenberger

TUNNELING AND STM CHARACTERIZATIONOF HTSC CUPRATES

595

Fig. 25. Differential conductance spectra measured as a function of position on an Y123 single-crystal sample at 4.2 K with applied field of 6 T: (a) between the vortex cores, (b) across a vortex core. The paths are 25 mm long. After Maggio-Aprile et al. (1995, fig. 3).

equation, that in d-wave superconductors quasiparticles within the vortex are in an extended state due to vanishing order parameter at /~ = (-t-g:v, 1 ± g1~ ) directions, and instead a differential conductance (d//dV) peak evolves at zero bias. In cuprate superconductors, Maggio-Aprile et al. (1995) have successfully observed the vortex lattice in Y123 for the first time. A fully oxygenated Y123 twinned single crystal grown in a novel crucible material, BaZrO3, was used in order to reduce contamination from the crucible during crystal growth. The low-temperature tunneling spectrum, shown in fig. 19 without magnetic field, changed as shown in fig. 25 under a field o f 6 T. The spectra varied while approaching the vortex core. The main 4-20 m e V conductance peaks decreased in magnitude and totally disappeared at the center o f the core. A t the same time, two new peaks grew at ±5.5 m e V The 27 m e V peak did not vanish completely, and a broad bump remained. However, no zero-bias p e a k was found at the center o f the vortex, in contrast to what is found in s-wave NbSe2. Maggio-Aprile et al. (1995) tentatively interpreted the new peaks at 5.5 m V as the lowest bound state for a vortex in s-wave superconductors. Here, the lowest bound energy is given approximately by E = z~/(~pF~) , where Pv and ~ are the Fermi momentum and the coherence length, respectively, implying PF~ ~ 1, the extreme quantum limit. It is also advocated that higher bound states are missing on the core spectrum, because their excitation energies exceed the superconducting gap A. Theoretical models beyond the simple BCS mechanism are discussed to elucidate the inconsistency between the tunneling results inside the vortex core and other data

596

H. SUGAWARA et al. ,0

2.5

....

i ....

i ....

i ....

i ....

~A~ k ".'~,lk2;x, -

I

i ....

~

2.0

.......

1.5

.......

i ....

i ....

,~ I~ "~,.A.~

i ....

1 ....

i ....

1,,,

(a) F~

I

I

" K~i

,~ . .

4.2K

!

0.5

T=83.0K (underdoped) , . . l

35 •

. . . .

I . . . .

I . . . .

'. ! . . . .

I . . . .

..': i . . . .

I . . . .

I . . . .

I . . . .

I . . . .

t , . .

' " ' .... ' .... ' .... ' .... ' .... ' .... ' .... ' .... ' .... ' .... ' " ' II (b) II

3.0

!

!

I

I

' 1.0 0.5

98.6K

/L,, '..

T¢:74.3 K (overdoped) ..,I

-300

....

I ....

-200

4.2 K

......

-....!

i ....

1 ....

-100

i ....

*-~-"v'~

i . . . , 1 . . , . I . , .

0

100

. .I . .

I ....

200

.] I ....

t

300

Vs.mpl~[mV] Fig. 26. Tunneling spectra ofBi2212 in the (a) underdoped and (b) overdoped regions. The spectra at 4.2 K were measured under a 6 T magnetic field at the center of the vortex core (solid line) and between vortices (dashed line). The spectra above To displayed as solid lines were measured in zero field, and reflect the normal-state psendogap structure. The dotted curves were obtained by numerically smearing the spectra measured at the center of the cores at 4.2 K and 6 T to the temperature(s) indicated for each doping level. They illustrate the striking resemblance of the vortex-core pseudogap and the normal-state pseudogap. For clarity, the 4.2 K curves are offset by 1.0 Gt72 ~ and the 88.7K curves by 0.4 G£2-l. After Renner et al. (1998b, fig. 3).

supporting d-wave. M o r i y a et al. (1997) claimed that the analysis from the Eilenberger equation is invalid in the case o f Y123 with its large value o f P F ~ , and that a calculation based on a more strict m o d e l ( B o g o l i u b o v - d e G e n n e s equation) o f d-wave superconductivity demonstrated that inner vortex excitation without a clear zero-bias peak. O n the other hand, Ogata a n d his group ( H i m e d a et al. 1997) theoretically showed from a calculation based on the t - J m o d e l that an s-wave c o m p o n e n t o f the order parameter i n d u c e d w i t h i n a vortex leads to the splitting o f the zero-bias peak;

TUNNELINGAND STM CHARACTERIZATIONOF HTSC CUPRATES

597

hence, the splitting of the peak structures indicates the admixure of d-wave and s-wave symmetries. Renner and coworkers (Renner et al. 1998b) have succeeded in observing vortex cores in under- and overdoped Bi2212 by STM. At the vortex core, they found a pseudogap-like feature without quasiparticle bound state. The inner-core gap structure at low temperature, as shown in fig. 26, is similar to the pseudogap above Tc, suggesting that both gap structures reflect the same normal state containing incoherent Cooper pairs. The missing bound states in Bi2212, in contrast to the Y123 case, suggest that the relatively larger gap in Bi2212 pushes up the first bound states E±1/2 beyond the superconducting gap.

6. Phase-sensitive m e a s u r e m e n t

While the quasiparticle tunneling process into superconductors is considered to be insensitive to the phase of the quasiparticle, several theoretical investigations have revealed that it is sensitive when the order parameter is anisotropic, in this section, two kinds of approaches to determine the symmetry order parameter of HTSC are reviewed. The first kind of approaches is to investigate the differential conductance (d//dV) spectrum with various geometries in SIN or SIS junctions. The other type is to measure the dependence of the Josephson critical current Jc as a function of applied magnetic field. 6.1. Dependence on surface orientations

6.1.1. In-plane anisotropy The group of Kane and Ng (Kane et al. 1994, Kane and Ng 1996) have adopted an unconventional junction geometry to observe the in-plane gap energy of Bi2212. Here, an edge of a single crystal faced in close proximity a normal-metal counter-electrode or a matching edge of the same crystal. The angular dependence of the energy gap was completely mapped by assuming fourfold symmetry with an experimental uncertainty of about 8 mV. The in-plane gap anisotropy was clearly observed with a minimum value of 2 0 m V along the CuO2 bond. The maximum gap direction was 45 ° away, with a value of about 40 mV The gap variation was larger than the experimental uncertainty of about 8 meV. It is not clear why there was a 45 ° difference between tunneling and photoemission results. No direct evidence of a nodal line was observed in the angular dependence.

6.1.2. Surface orientations and zero-bias anomaly The tunneling process across a SIN (superconductor-insulator-normal metal) junction is more complicated than across a NIN (normal metal-insulator-normal metal) junction described in sec. 3.2. An electron reaching an insulator (I) from a normal metal (N) takes either one of four elementary processes, i.e., (1) reflection as an electron

598

H. SUGAWARAet al.

(normal reflection), (2) reflection as a hole after the interaction with an electron-like quasiparticle (Andreev reflection), (3) transmission as an electron-like quasiparticle, and (4) transmission as a hole-like quasiparticle. The tunneling current should include these processes, and the differential conductance is given by the Blonder-TinkhamKlapwijk (BTK) formula (Blonder et al. 1982). The tunneling conductance for anisotropic superconductors in general was calculated by Tanaka and Kashiwaya (Y. Tanaka and Kashiwaya 1995, Kashiwaya et al. 1995b, 1996). Their formula reproduces the Blonder-Tinkham-Klapwijk (BTK) formula for s-wave superconductors when the barrier height is sufficiently high, i.e., tunneling conductance approximates the quasiparticle density of states. On the other hand, the tunneling conductance for d-wave superconductors is sensitive to the phase of the quasiparticles. In the case of ttumeling in an oblique direction, the phase difference of pair potentials induces zero-energy states (ZES's) at zero energy of quasiparticles (Hu 1994), which create a conductance peak at zero bias (ZBCP). The ZBCP is not expected for s-wave superconductors of any orientation, nor for the basal (ab-) plane of d-wave superconductors. Hence, the appearance of ZBCP for tunneling in transverse direction is considered to be support for an anisotropic superconductor, in particular, dx2 y2-wave. Tumleling experiments have been performed for several surface orientations; some results were in agreement with the theoretical predictions above, others were in contradiction. Observations of the ZBCP are reported on (110) surfaces of Y123 (Alff et al. 1997, Wei et al. 1998), and also on (100) surfaces of Y123 (Kashiwaya et al. 1995a,b, Suzuki et al. 1999, Covington et al. 1996b, Wei et al. 1998) and (100) surfaces of La214 (S. Tanaka et al. 1995a,b). However, the ZBCP was not observed on (110) surfaces of Y123 by Nantoh eta!. (1995), Kane et al. (1994) and Suzuki et al. (1999), and not observed on (110) surfaces of La214 by S. Tanaka et al. (1995a,b). Tanuma et al. (1998) calculated the tunneling density of states on the uneven surface of d-wave superconductors and showed that their calculation reproduced various types of anomalous features observed in the actual tunneling experiments, such as ZBCP, double-peak and multiple-dip structures, and a suppressed superconducting gap. They claimed that the wide variety of experimental data showing these features are natural for anisotropic superconductors, and hence, should not be rejected as unidentified spectra observed on degraded surface or in "bad junctions" (Kashiwaya et al. 1994a). 6.2. Josephson tunneling under magnetic field 6.2.1. Josephson tunneling across Y123/Pb junction in the ab-plane Several measurements have been performed to study the dependence of Josephson critical current flowing within the CuO2 planes on applied magnetic field (i.e. Fraunhofer pattern) for a junction made of Y123 and Pb. These measurements have been proposed to determine whether the order parameter of Y123 has d-wave symmetry or not; the diffraction pattern should be dependent on the configuration of the junction, as shown in fig. 27. For a junction between two s-wave superconductors, or for an "edge" junction between d-wave and s-wave superconductors (fig. 27a), the diffraction pattern is normal

T U N N E L I N G A N D S T M C H A R A C T E R I Z A T I O N OF H T S C C U P R A T E S

599

1

(a)

-6

-4

-2

2

4

0 2 @/oo o

4

@/@0 1

(b)

0 -6

%

-4

-2

I

1

d-wave

0 -6

-4

-2

0

@/@o

2

4

Fig. 27. (a) Fraunhofer diffraction pattern for the critical current vs magnetic flux threading the junction barrier that is characteristic of a single Josephson tunnel junction. (b) Diffraction patterns expected for a single corner junction with s-wave (isotropic or anisotropic) and d~2 y2 pairing symmetry. After Wollman et al. (1995, fig. 1).

Fraunhofer-like. The Josephson critical current oscillates according to the following equation: I t ( q 0 = Ic sin (~q~/q~o)

~q~/q~o

'

(4)

where q~o is the magnetic flux quantum. On the other hand, the corresponding expression for a "corner" junction (fig. 27b) between the s-wave and d-wave superconductors is sin 2 ( ~ / 2 ~ o ) i~(q,) = Ic

~q~/2 q~0

(5)

Wollman et al. (1993, 1995) reported detailed measurements o f the magnetic field dependence o f the Josephson critical current for Y123/Au/Pb junctions formed on Y123 crystals. The observed results o f the dependences for the edge junction and the corner junction are as shown in fig. 28. Double peak structures were seen for the corner junction, as expected from equation (5), indicating that Y123 is a d-wave superconductor.

600

H. SUGAWARAet al.

100 80 60

~

'

'

(a)

40 "F--

o

20 0

-200

-100 0 100 200 Applied magneticfield (mG)

200 ~

i

(b)

150 100 50

c) 0

-1000 -500 0 500 1000 Applied magneticfield (mG)

Fig. 28. Measured critical current vs. applied magnetic field for (a) an "edge" YI23/Pb junction and (b) a "corner"junction including ab faces. AfterWollmanet al. (1995, fig. 3).

When a trapped flux was present, the double peak structures were found to be deformed asymmetrically, which was attributed to a nonuniform magnetic field across the junction barrier. Corner-junction experiments by other groups (Brawner and Ott 1994, Mathai et al. 1995, Miller et al. 1995) and tricrystal ring experiments (Tsuei et al. 1994, Kirtley et al. 1995) also provide evidence for d-wave pairing symmetry. 6.2.2. Josephson tunneling across" Y123/Pb junction along the c-axis Contrary to the results of phase-sensitive measurements involving current flowing within the CuO2 planes, measurements of Josephson tunneling currents perpendicular to the CuO2 planes between heavily twinned Y123 and a conventional s-wave superconductor (Sun et al. 1994, 1996, Katz et al. 1995, Lesueur et al. 1997, Kleiner et al. 1996) resulted in the observation of finite supercurrents, suggesting a significant s-wave component. In order to elucidate this contradiction, Dynes and his group (Kouznetsov et al. 1997) measured the Josephson-pair tunneling current along the c-axis from conventional Pb superconductors across thin insulating layers into crystals or thin films of Y123. Pb is deposited across a single twin boundary ofY123 crystal, as shown in fig. 29. A c-axis Josephson current is sensitive to the net phase difference with any d-wave component. The current flowing into the Pb electrode is the sum of the tunneling currents

TUNNELING AND STM CHARACTERIZATIONOF HTSC CUPRATES

601

Fig. 29. (a) Schematic diagram of a Pb/Y123 junction (hatched region) grown across a twin boundary (dashed line). For the Y123 order parameter shown, the s-wave component changes sign across the boundary. (b) Nemarsky microscope photograph of junction. The twin boundary runs vertically through the center of file photograph; the Pb strip is the horizontal white region. After Kouznetsov et al. (1997, fig 1).

from each o f the twin domains. I f Y123 were predominantly d-wave, the phase o f the dx2 y2-wave order parameter should be maintained across the twin boundary, and the resulting wavefunction can be written as sums o f a dx2_y2-wave state and an s-wave state, specifically as d + s and d - s . Under the same assumption, the Josephson critical current is expected to be strongly dependent on the direction o f the magnetic field applied within the plane because the s-component would change sign across the twin boundary. Figure 30 shows the Josephson critical current measured under external fields along different directions. With the field perpendicular to the twin boundary (q~ ~ 4-90 ° in the figure), the plot o f the critical current exhibited the same Fraunhofer pattern as observed in an ordinary junction, except for a lower peak current. With the field parallel to the boundary (q~ = 0), a dip rather than a peak appeared at zero field (B = 0), and the m a x i m u m current occurred at a field value corresponding to half-integer quantum flux, ½~0. These observations indicate that the flux cancels the phase difference between the two domains and causes a current flow in the same direction. These observations indicate that Y123 favors d-wave symmetry with some s-wave admixture. 6.2.3. Splitting of zero-bias conductance peak in zero magnetic field The conductance peak near zero bias voltage (ZBCP) has been observed when tunneling takes place into ab-oriented Y123 thin films (Geerk et al. 1988, Lesueur et al. 1992,

H. SUGAWARA et al.

602 300

300

250

250

200

200

o~ 150

150

E

0

o 100

100

50

50

0

0 -0.2

0 magnetic field (mT)

0.2

-0.2

0

0.2

magnetic field (mT)

Fig. 30. The critical current I,.(B) measured for ten values of angle ~ relative to the twin boundary (TB) in Pb/Y123 junction. Successive plots are offset vertically by 30p~A. Center inset: Computed values of I,./I o vs ~/~0 for a junction with sign reversal of the s-wave component. Successive plots are offset vertically by 0.3. After Kouznetsov et al. (1997, fig. 3).

Covington et al. 1996a). It was reported that the zero-bias conductance peak split when a magnetic field is applied (Lesueur et al. 1992); however, Covington et al. (1997) observed that the conductance peak split even when no magnetic field was applied, once the temperature was lowered below about 7 K, as shown in fig. 31. In the experiment of Covington et al. (1997), Y123/insulator/Cu planar tunneling junctions were fabricated on Y123 ab-oriented thin films. It was found that the zerobias conductance G(0)= (d//dV)l v-0 increased with decreasing temperature around 10 K, below which it decreased. The applied magnetic field induced further splitting that grew nonlinearly with increasing field. The field dependence of G(0) showed significant discrepancy with the calculation based on spin-flip scattering due to the magnetic impurities proposed by Lesueur et al. (1992), as shown in the inset of fig. 31. On the other hand, Covington et al. (1997) have claimed that the splitting of the ZBCP indicates a broken time-reversal symmetry state at the surface of Y123, where the Andreev bound state shifts to finite energy, resulting in a split of ZBCP (Fogelstr6m et al. 1997, Matsumoto and Shiba 1996). The phase diagram for the broken time-reversal symmetry state calculated by Fogelstr6m et al. (1997) was in good agreement with their results. The gap-like feature and ZBCP were observed at equal strength in the

TUNNELING AND STM CHARACTERIZATIONOF HTSC CUPRATES

0.28

vo

024

0

0.20

i

l

603

"r

4 2 K

-20

--1.5K

-1'0

&

1'0

20

Voltage (mY) Fig. 31. Temperature-dependent conductance data from a Y123/Cu turmel junction. Voltage is defined to be the voltage of Cu with respect to Y123. The ZBCP is observed to split in zero magnetic field at 1.5 K. Inset: zero-bias conductance, G(0), versus temperature for the samejunction, also in zero field. The downturn in G(0) below 10 K is in contrast to the G(0) ~ ln(T) behavior expected from spin-flip scattering proposed by Lesueur et al. (1992), which is indicated by the dotted line. After Covington et al. (1997, fig. 1). tunneling conductance for (100)- and (1 lO)-oriented films, which indicates comparable pair breaking for both orientations; this suggests that s-wave is the most plausible subdominant gap symmetry.

7. Summary Recent achievements in the tunneling study for cuprate superconductors are summarized as follows: (1) In the high-temperature oxide superconductors, in particular Bi2212 and T12223, alternating stacking o f superconducting and blocking layers is recognized as intrinsic Josephson junctions. Tunneling currents flowing through small junctions have been extensively investigated, indicating Josephson plasma excitation, characteristic for cuprate superconductors. (2) Vacuum tunneling has been performed utilizing scanning tunneling microscope at low temperature. Removing the topmost surface by cleavage in vacuum or an inert gas atmosphere prior to tunneling measurements has been demonstrated to be indispensable to observe intrinsic superconducting gap structure in Bi2212. The reported superconducting gap spectra have been analyzed in the framework of d-wave superconductivity. It has been found that results can be, in principle, described in terms o f the d-wave order parameter. (3) Hole doping and temperature dependence have been further investigated. Gap-like structures have been reported for underdoped Bi2212 samples above Tc. These

604

H. SUGAWARA et al.

gap-like structures should be understood in accordance with the "pseudogap" observed in other types of measurements. (4) Using STM, inner vortex structures of the differential conductance were successfully visualized for Bi2212 and Y123 crystals. The inner vortex excitation spectra are found to be quite different from those expected from a simple d-wave mechanism. (5) Phase-sensitive measurements have been carried out for Y123 crystals with different orientations. Josephson tunneling at the interface between conventional and hightemperature superconductors has also been investigated. The observed tunneling spectra are highly dependent on geometry; some results agree with theoretical predictions from d-wave symmetry, while others do not. Better theoretical models, including the admixture of d-wave and s-wave order parameters, have been proposed.

Acknowledgement We acknowledge the Japan Science and Technology (JST) corporation for supporting this work.

References Atff, L., H. Takashima, S. Kashiwaya, N. Terada, H. lhara, Y. Tanaka, M. Koyanagi and K. Kajimura, 1997, Phys. Rev. B 55, 14757. Alloul, E, T. Otmo and R Mendels, 1989, Phys. Rev. Lett. 63, 1700. Anderson, P.V¢, 1987, Science 235, 1196. Anderson, P.W., and Z. Zou, 1988, Phys. Re,/. Lett. 60, 132. Balatsky, A.V., and M.I. Salkola, 1996, Phys. Rev. Lett. 76, 2386. Balzarotti, A., M. de Crescenzi, E Patella, A. Sgarlata, E Paroli, G. Balestrino and M. Marinelli, 1991, Phys. Rev. B 43, 11500. Barbiellini, B., O. Fischer, M. Peter, Ch. Renner and M. Wegner, 1994, Physica C 220, 55. Basov, D.N., R. Liang, B. Dabrowski, D.A. Bonn, WN. Hardy and T. Timusk, 1996, Phys. Rev. Lett. 77, 4090. Blonder, G., M. Tinkham and T. Klapwijk, 1982, Phys. Rev. B 25, 4515. Brawner, D.A., and H.R. Ott, 1994, Phys. Rev. B 50, 6530. Broekholt, M., T. Bollmeier, L. Buschmann, M. Fleuster and G. Gfintherodt, 1992, Physica C 198, 33. Caroli, C., RG. de Gennes and J. Matricon, 1964, Phys. Lett. 9, 307.

Chang, A., Z.Y. Rong, Yu.M. Ivanchenko, E Lu and E.L. Wolf, 1992, Phys. Rev. B 46, 5692. Covington, M., R. Scheuerer, K. Bloom and L. Greene, 1996a, Appl. Phys. Lett. 68, 1717. Covington, M., E Xu, C. Feldmann and L. Greene, 1996b, Czech. J. Phys. 46, 1341. Covington, M., M. Aprile, E. Paraoanu and L. Greene, 1997, Phys. Rev. Lett. 79, 277. Ding, H., M.R. Norman, J.C. Campuzano, M. Randeria, A. Bellman, Z Yokoya, T. Takahashi, T. Mochiku and K. Kadowaki, 1996, Phys. Rev. B 54, 9678. Edwards, H.L., J.T. Markert and A.L. de Lozanne, 1992, Phys. Rev. Lett. 69, 296% Edwards, H.L., A.L. Barr, J.T. Markert and A.L. de Lozanne, 1994a, Phys. Rev. Lett. 73, 1154. Edwards, H.L., J.T. Markert and A.L. de Lozamle, 1994b, J. Vac. Sei. Technol. B 12, 1886. Edwards, H.L., D.J. Derro, A.L. Barr, J.T. Markert and A.L. de Lozanne, 1995, Plays. Rev. Lett. 75, 1387. Emery, V.J., and S.A. Kivelson, 1995, Phys. Rev. Lett. 74, 3253. Flensberg, K., E Hedeghrd and M. Pedersen, 1988, Phys. Rev. B 38, 841. Fogelstr6m, M., D. Rainer and J.A. Sauls, 1997, Phys. Rev. Lett. 79, 281. Geerk, J., X. Xi and G. Linker, 1988, Z. Phys. B 73, 329.

TUNNELING AND STM CHARACTERIZATION OF HTSC CUPRATES Hancotte, H., R. Deltour, D. Davydov, A. Jansen and E Wyder, 1997, Phys. Rev. B 55, 3410. Harmer, M., C. Fincher and B. Parkinson, 1991, J. Appl. Phys. 70, 2760. Hasegawa, T., M. Nantoh, H. lkuta and K. Kitazawa, 1991a, Physica C 185-189, 1743. Hasegawa, T., M. Nantoh and K. Kitazawa, 1991b, Jpn. J. Appl. Phys. 30, L276. Hasegawa, T., H. lkuta and K. Kitazawa, 1992a, in: Physical Properties of High Temperatm'e Superconductors. Vol. III, ed. D. Ginsberg (World Scientific, Singapore) p. 525. Hasegawa, T., A. Takagi, H. Ikuta, M. Kawasaki, H. Koinuma and K. Kitazawa, 1992b, J. Phys. Chem. Solids 53, 1643. Hasegawa, T., M. Nantoh, S. Heike, A. Takagi, M. Ogino, M. Kawasaki, H. Koinuma and K. Kitazawa, 1993, J. Phys. Chem. Solids 54, 1351. Hawley, M., I.D. Raistrick, J.G. Beery and R.J. Houlton, 1991, Science 251, 1587. Helm, C., C. Preis, E Forsthofer, J. Keller, K. Schlenga, R. Kleiner and R Mfiller, 1997, Phys. Rev. Lett. 79, 737. Hess, H.E, R.B. Robinson, R.C. Dynes, J.M. Valles Jr and J.V Waszczak, 1989, Phys. Rev. Lett. 62, 214. Himeda, A., M. Ogata, Y. Tanuma and S. Kashiwaya, 1997, J. Phys. Soc. Jpn. 66, 3367. Homes, C.C., T. Timusk, R. Liang, D.A. Bonn and W.N. Hardy, 1993, Phys. Rev. Lett. 71, 1645. Honcotte, H., D. Davydov, M. Ye and R. Deltour, 1995, Physica B 204, 206. Hu, C., 1994, Phys. Rev. Lett. 72, 1526. Hwang, H.Y., B. Batlogg, H. Takagi, H.L. Kao, J. Kwo, R.J. Cava, J.J. Krajewski and W.E Peck Jr, 1994, Phys. Rcv. Lett. 72, 2636. Ichimura, K., and K. Nomura, 1993, J. Phys. Soc. Jpn. 62, 3661. Ichioka, M., N. Hayashi, N. Enomoto and K. Machida, 1996, Phys. Rev. B 53, 15316. Ikuta, H., A. Maeda and K. Uchinokura, 1988, Jpn. J. Appl. Phys. 27, L1039. Ishida, K., Y. Kitaoka, N. Ogata, T. Kamino, K. Asayama, J.R. Cooper and N. Athanassopoulou, 1993, J. Phys. Soc. Japan 62, 2803. Johnston, D., 1989, Phys. Rev. Lett. 62, 957. gane, J., and K.-W. Ng, 1996, Phys. Rev. B 53, 2819. Kane, J., Q. Chen, K.-W. Ng and H.-J. Tao, 1994, Plays. Rev. Lett. 72, 128. Kane, J., K.-W. Ng and D. Moecher, 1998, Physica C 294, 176.

605

Kaneko, S., N. Nishida, T. Mochiku and K. Kadowaki, 1998, Physica C 298, 105. Kashiwaya, S., M. Koyanagi, H. Takashima, M. Matsuda and K. Kajimura, 1994a, in: Advances in Superconductivity. Vol. VI, eds T. Fujita and Y. Shiohara (Springer, Tokyo) p. 73. Kashiwaya, S., M. Koyanagu, M. Matsuda and K. Kajimura, I994b, Physica B 194 196, 2119. Kashiwaya, S., Y. Tanaka, M. Koyanagi, H. Takashima and K. Kajimura, 1995a, J. Phys. Chem. Solids 56, 1721. Kashiwaya, S., Y. Tanaka, M. Koyanagi, H. Takashima and K. Kajimura, 1995b, Phys. Rev. B 51, 1350. Kashiwaya, S., Y. Tanaka, M. Koyanagi and K. Kajimura, 1996, Phys. Rev. B 53, 2667. Katz, A.S., A.G. Sun, R.C. Dynes and K. Char, 1995, Appl. Phys. Lett. 66, 105. Kawasaki, M., and M. Nantoh, 1994, MRS Bulletin XIX, 33. Kendziora, C., and A. Rosenberg, 1994, Physica C 235 240, 1t21. Kirk, M., J. Nogami, A. Baski, D.B. Mitzi, A. Kapitulnik, T.H. Geballe and C. Quate, 1988, Science 242, 1673. Kirtley, J.R., and D.J. Scalapino, 1990, Phys. Rev. Lett. 65, 798. Kirtley, J.R., C.C. Tsuei, J.Z. Sun, C.C. Chi, L.S. YuJahnes, A. Gupta, M. Rupp and M.B. Ketchen, 1995, Nature 373, 225. Kitazawa, K., H. Sugawara and T. Hasegawa, 1996, Physica C 263, 214. Kleiner, R., and P. Mfiller, 1994, Phys. Rev. B 49, 1327. Kleiner, R., E Steinmeyer, G. Kunkel and P. Mfiller, 1992, Phys. Rev. Lett. 68, 2394. Kleiner, R., A.S. Katz, A.G. Sun, R. Summer, D.A. Gajewski, S.H. Han, S.I. Woods, E. Dantsker, B. Chen, K. Char, M.B. Maple, R.C. Dynes and J. Clarke, 1996, Phys. Rev. Lett. 76, 2161. Kouznetsov, K.A., A.G. Sun, B. Chen, A.S. Katz, S.R. Bahcall, J. Clarke, R.C. Dynes, D.A. Gajewski, S.H. Han, M.B. Maple, J. Giapintzakis, J.-T. Kim and D.M. Ginsberg, 1997, Phys. Rev. Lett. 79, 3050. Lang, H., T. Frey and H.-J. Gi.intherodt, 1991, Europhys. Lett. 15, 667. Lesueur, J., L. Greene, W. Feldmarm and I. Inam, 1992, Physica C 191, 325. Lesueur, J., M. Apriti, A. Goulan, T. Horton and L. Dumoulin, 1997, Phys. Rev. B 55, 3398. Liu, J., Y. Li and C. Lieber, 1994, Phys. Rev. B 49, 6234.

606

H. SUGAWARA et al.

Liu, J.-X., J.-C. Wan, A. Goldman, Y.C. Chang and P.-Z. Jiang, 1991, Phys. Rev. Lett. 67, 2195. Loeser, A.G., Z.-X. Shen, D.S. Dessau, C.-H. Park, R Fom-nier and A. Kapitulnik, 1996, Science 273, 325. Loram, J.W., K.A. Mirza, J.R. Cooper and W.Y. Liang, 1993, Phys. Rev. Lett. 71, 1740. Loram, J.W., K.A. Mirza, J.R. Cooper, W.Y. Liang and J.M. Wade, 1994, J. Supercond. 7, 243. Maggio-Aprile, I., Ch. Renner, A. Erb, E. Walker and O. Fischer, 1995, Phys. Rev. Lett. 75, 2754. Maid, K., Y. Stm and H. Won, 1996, Czech. J. Phys. 46(Suppl. 6), 3151. Mallet, P., D. Roditehev, W. Sacks, D. D6fourneau and J. Klein, 1996, Phys. Rev. B 54, 13324. Manabe, C., M. Oda and M. Ido, 1994, Physica C 235-240, 797. Manabe, C., M. Oda and M. Ido, 1997, J. Phys. Soc. Jpn. 66, 1776. Mandrus, D., J. Hartge, C. Kendziora, L. Mihaly and L. Forro, 1993, Europhys. Lett. 22, 199. Mathai, A., Y. Gim, R. Black, A. Amar and E Wellstood, 1995, Phys. Rev. Lett. 74, 4523. Matsuda, A., S. Sugita and T. Watanabe, 1999, Phys. Rev. B 60, 1377. Matsumoto, M., and H. Shiba, 1996, J. Phys. Soc. Jpn. 65, 2194. Matsumoto, T., S. Choopun and T. Kawai, 1995, Phys. Rev. B 52, 591. Matsuura, S., T. Taneda, W Yamaguchi, H. Sugawara, T. Hasegawa and K. Kitazawa, 1998, Physica C 300, 26. Miller, J., Q. Ying, z. Zou, N. Fan, J. Xu, M. Davis and J. Wolfe, 1995, Phys. Rev. Lett. 74, 2347. Miller, T., M. McElfresh and R. Reifenberger, 1993, Phys. Rev. B 48, 7499. Miyakawa, N., P. Guptasarma, J. Zasadzinski, D. Hinks and K. Gray, 1998, Phys. Rev. Lett. 80, 157. Momono, N., R. Dipasupil, H. Ishignro, S. Saigo, T. Nakano, M. Oda and M. Ido, 1999, Physica C 31%318, 603. Moreland, J., R Rice, S. Russek, B. Jeanneret, A. Roshko, R. Ono and D. Rudman, 1991, Appl. Phys. Lett. 59, 3039. Moriya, Y., M. Kohmoto and K. Mald, 1997, Phys. Rev. Lett. 78, 4841. Muenchausen, R., M. Hawley, S.R. Foltyn, X. Wu, R. Dye and E Garzon, 1992, Physica C 199, 445. Murakami, H., and R. Aoid, 1995, J. Phys. Soc. Jpn. 64, 1287.

Nagaosa, N., and EA. Lee, 1990, Phys. Rev. Lett. 64, 2450. Nakamura, q2, and S. Uehida, 1993, Phys. Rev.B 47, 8369. Nakano, T., M. Oda, C. Manabe, N. Momono, Y. Miura and M. ldo, 1994, Phys. Rev. B 49, 16000. Nantoh, M., S. Heike, H. Ikuta, T. Hasegawa and K. Kitazawa, 1991, Physica C 185-189, 861. Nantoh, M., T. Hasegawa, W. Yamaguchi, A. Takagi, M. Ogino, M. Kawasaki, J. Gong, H. Koinuma and K. Kitazawa, 1994, J. Appt. Phys. 75, 5227. Nantoh, M., M. Kawasaki, T. Hasegawa, K. Fujito, W. Yamaguchi, H. Koinuma and K. Kitazawa, 1995, Physica C 242, 277. Narlikar, A., P. Durra, S. Samanta, O. Srivastava, P. Ramasamy, S. Sabharwal, M. Gupta and B. Padalia, 1992, J. Cryst. Growth 116, 37. Narlikar, A., S. Samanta, C. Changkang, H. Yongle, J. Hodby and B. Wanklyn, 1996, J. Cryst. Growth 158, 248. Nishikawa, T., J. Takeda and M. Sato, 1994, J. Phys. Soc. Jpn. 63, 1441. Oda, M., C. Manabe and M. Ido, 1996, Plays. Rev. B 53, 2253. Oda, M., K. Hoya, R. Kubota, C. Manabe, N. Momono, T. Nakano and M. Ido, 1997a, Physica C 281, 135. Oda, M., K. Hoya, R. Kubota, C. Manabe, N. Momono, T. Nakano and M. Ido, 1997b, Physica C 282-287, 1499. Panagopoulos, C., and T. Xiang, 1998, Phys. Rev. Lett. 81, 2336. Rennet, Ch., and O. Fischer, 1995, Phys. Rev. B 51, 9208. Renner, Ch., A. Kent, P. Niedermann, O. Fischer and E L6vy, 1991, Phys. Rev. Lett. 67, 1650. Renner, Ch., O. Fischer, A. Kent, D.B. Mitzi and A. Kapitulnik, 1994, Physica B 194, 1689. Renner, Ch., B. Revaz, J.-Y. Genoud, K. Kadowaki and O. Fischer, 1998a, Phys. Rev. Lett. 80, 149. Renner, Ch., B. Revaz, K. Kadowaid, I. Maggio-Aprile and O. Fischer, 1998b, Phys. Rev. Lett. 80, 3606. Rossat-Mignod, J., L.R Regnault, C. Vettier, P. Burlet, J.Y. Henry and G. Lapertot, 1991, Physica B 169, 58. Salkola, M.I., A.V Balatsky and D.J. Scalapino, 1996, Phys. Rev. Lett. 77, 1841. Schlenga, K., G. Hechtfischer, R. Kleiner, W. Walkenhorst, R Miiller, H. Johnson and E. Steinbeig, 1996, Phys. Re,/. Lett. 76, 4943. Schlenga, K., R. Kleiner, G. Hechtfischer, M. M6Ble, S. Schmitt, R Mtiller, C. Helm, C. Preis,

TUNNELING AND STM CHARACTERIZATION OF HTSC CUPRATES E Forsthofer, J. Keller, H. Johnson, M. Veith and E. Steinbeil3, 1998, Phys. Rev. B 57, 14518. Schlesinger, Z., 1994, Physica C 235-240, 49. Schopohl, N., and K. Maki, 1995, Phys. Rev. B 52, 490. Shih, C., R. Feensta'a, J. Kirtley and G. Chandrashekhar, 1988, Phys. Rev. B 40, 2682. Slakey, E, M.V. Klein, J.R Rice and D.M. Ginsberg, 1990, Phys. Rev. B 42, 2643. Sugawara, H., M. Nantoh, T. Hasegawa and K. Kitazawa, 1996, in: Studies on High Tcmperatm'e Superconductors, Vol. 20 (Nova Science Publishers, New York) pp. 1-39. Sun, A.G., D.A. Gajewski, M.B. Maple and R.C. Dynes, 1994, Phys. Rev. Lett. 72, 2267. Sun, A.G., A. Truscott, A.S. Katz, R.C. Dynes, B. Veal and C. Gu, 1996, Phys. Rev. B 54, 6734. Suzuki, K., K. Ichimura, K. Nomura and S. Takekawa, 1999, J. Phys.: Condens. Matter 11, 3133. Snzumma, Y., Y. Hasegawa and H. Fukuyama, 1988, J. Phys. Soc. Jpn. 57, 1278. Tajima, S., J. Schutzmann, S. Miyamoto, [. Terasaki, Y. Sato and R. Hauff, 1996, Phys. Rev. B 55, 6051. Takagi, H., B. Batlogg, H.L. Kao, J. Kwo, R.J. Cava, J.J. Krajewski and W.E Peck Jr, 1992, Phys. Rev. Lett. 69, 2975. Tallon, J.L., J.R. Cooper, R de Silva, G. Williams and J.W. Loram, 1995, Phys. Rev. Lett. 75, 4114. Tanabe, K., Y. Hidaka, S. Karimoto and M. Suzuki, 1996, Phys. Rev. B 53, 9348. Tanaka, S., E. Ueda, M. Sato, K. Tamasaku and S. Uchida, 1995a, J. Phys. Soc. Jpn. 64, 1476. Tanaka, S., E. Ueda, M. Sato, K. Tamasaku and S. Uchida, 1995b, J. Phys. Soc. Jpn. 65, 2212. Tanaka, Y., and S. Kashiwaya, 1995, Phys. Rev. Lett. 74, 3451. Tanuma, Y., Y. Tanaka, M. Yamashirn and S. Kashiwaya, 1998, Phys. Rev. B 57, 7977.

607

Tsai, J., I. Takeuchi, J. Fujita, S. Miura, T. Terashima and Y. Bando, 1989, Physica C 157, 537. Tsuei, C.C., J.R. Kirtley, C.C. Chi, L.S. Yu-Jahnes, A. Gupta, T. Shaw, J.Z. Sun and M.B. Ketchen, 1994, Phys. Rev. Lett. 73, 593. Ueda, E., S. Tanaka, M. Sato, K. Tamasaku and S. Uchida, 1995, Physica C 249, 181. Valles Jr, J.M., R.C. Dynes, A.M. Cucolo, M. Gurvitch, L.E Schneemeyer, J.R Garno and J.V Waszczak, 1991, Phys. Rev. B 44, 11986. Varma, C.M., RB. Littlewood, S. Schmitt-Rink, E. Abrahams and A.E. Ruckenstein, 1989, Phys. Rev. Lett. 63, 1996. Wang, C., B. Giambattista, C. Slough, R. Coleman and M. Subramanian, 1990, Phys. Rev. B 42, 8890. Wei, J., N.-C. Yeh, D. Garrigus and M. Strasik, 1998, Phys. Rev. Lett. 81, 2542. Wolf, E.L., A. Chang, Z.Y. Rong, Yu.M. Ivanchenko and F. Lu, 1994, J. Supercond. 7, 355. Wollman, D.A., D.J. van Harlingen, W.C. Lee, D.M. Ginsberg and A.J. Leggett, 1993, Phys. Rev. Lett. 71, 2134, Wollman, D.A., D.J. van Harlingen, J. Giapintzakis and D.M. Ginsberg, 1995, Phys. Rev. Lett. 74, 797. Won, H., and K. Maki, 1996, J. Phys. l 6, 2317. Wu, X., Z. Zhang, Y. Wang and C. Lieber, 1990, Science 248, 1211. Wu, X., Y. Wang, Z. Zhang and C. Lieber, 1991, Phys. Rev. B 43, 8729. Yasuoka, H., T. Imai and T. Shimizu, 1989, in: Strong Correlations and Superconductivity, eds H. Fukuyama, S. Maekawa and A.P. Malozemoff (Springer, Berlin) p. 254. Yurgens, A., D. Winkler, N. Zavaritsky and T. Claeson, 1996, Phys. Rev. B 53, 8887.

SUBJECT INDEX BVS (Bond Valence Sum) method 30, 93-99 application to 123-O~ 30-33, 93 BaxK~ ~BiO3 535 BaO surface 403 BaPrO3 344 Ba surface component 403 band dispersions 398 Bean model t96-198, 206, 239 BiO layer 581 BiO plane 577 Bi2Sr2(Ca062Y0.38)Cu208 540 Bi2Sr2CaCu20x 264, 266 Bi2Sr2CaCu2Q 260, 262, 271,272, 439, 451, 453, 459, 462, 467, 468, 475, 477, 485, 527, 528, 532-534, 536, 539-541, 543-545, 548, 549, 551,566, 568 (001) surface 577-582 - Bi-O layer 577 clearing, etching 578-581 - Cu-O plane 577 doping 588 - overdoped 589, 591 oxygen content 589 underdoped 568, 589, 591 Bi2Sr2CaCu2Q + ,7 259, 260, 269, 418, 496 Bi2Sr~CuQ. 255, 260 Bi2Sh_.La.CuO6 492 bilayer splitting 415, 418, 425, 432 Blonder Tinkham-Klapwijk (BTK) tbrmula 598 Bond Valence Sum, s e e BVS method bonding ~r state 414 Bose Einstein condensation 512 Bose thermal factor 518 bosonic spectral functions, s e e A ( g 2 ) bound states 595, 597 break junction 567

ARPES (angle-resolved photoelectron spectroscopy) 391-432, 467, 469, 47t, 574 c-axis dispersion 415, 432 - lineshapes 408 ASYNNNI model for 123-Ox 87-93 calculation of phase diagram 87 abnormal metallic region 573 abnormal normal region 570 Abrikosov-Gorkov 316, 326 Andreev reflection 598 angle-resolved photoelectron spectroscopy, s e e ARPES angle-resolved photoemission 399 anisotropy 361 anti-chain (05) site occupancies in 123-Ox 61 anti-Stokes scattering 514 antibonding (7* state 414 antiferromagnetic currelations 418, 419 antiferromagnetic exchange constants 367 antiferromagnetic Fermi liquid 535, 539 antiferromagnetic insulator 511 antiferromagnetic neutron peaks 368 antiferromagnetic ordering 355, 358 antiferromagnetic region 574 antifermmagnetism 282, 294 in La2CuO4 284 antiferromagnets 381,522 A(£2) 470 apical oxygen 401 splitting of 73 artificial pinning centers 212, 213, 221-227, 231, 234, 235, 241 asymmetric next-nearest neighbor Ising model, s e e ASYNNNI model

-

-

-

-

-

-

-

-

BAO synthesis of 123-O,. 13 - bond length of samples 56 lattice parameters of samples 44 BCS (Bardeen-Cooper-Schrieffer) model 484, 542 -

473,

-

677

CAR synthesis of 123-O., 13 lattice parameters of smnples 39-43 CT gap, s e e charge transfer gap

678

SUBJECT INDEX

Ca doping 356 - in 123-Ox 163-165 - effective charge screening of Ca 2+ in Ca-123-Q. 151 localization of the holes 163 Ca-123-O~. I51-162 Ca percolation 160 - EXAFS investigations 159-162 magnetic characterization 156-159 Raman characterization 154-156 stl~ctural characterization 151-154 Ca05Sr0.~RuO 3 535 canted magnetic moments 369, 380 carbonate content of 123-Ox, mass-spectrometric determination 20-22 CeBa2Cu30 x 354 (Ce I .,.Rx)Ru2 316 chain copper 401 chain layer 331,332 chain magnetism 343 chain oxygen 401 chain sheet 422, 432 chain spin moment 341 charge ordering 380 charge transfer (CT) gap 446, 449, 511, 525 Chevrel phase 316 clean-limit superconductor 484 cleavage 585 cluster calculations 362 coated conductor 196, 236, 242 coherence length 393, 565 collective mode 41 meV peak 472 collinear magnetic structure 319 commensurate antiferromagnets 318 competition between superexchange and RKKY interaction 368, 380 complex conductivity 442 conductivity 442--472 - along C ~ O chains 460-462 - in-plane 451M-53, 456-460 - interplane 462-466 contributions to the specific heat 352 354 Cooper pair 587 copper-oxygen bonding 318 "comer" junction 599 correlation length 376 Coulomb exchange interaction 329 Coulomb screening 519, 520, 542, 546, 547, 549 coupling constants 361 -

-

-

-

critical current 568 critical current density 188, 192, 193, 195-i99, 201, 202, 204, 206, 209, 212 214, 221,225, 227, 228, 231,232, 234, 239, 241 crystal momentum 398 crystal structure 1-2-3 332 - 2-4-7 332 2-4-8 332 - Y123 401 crystalline electric field (CEF) 352 level schemes 360 parameters 360, 362 spectroscopy 368 splitting 381 Cu O bonds in 123-Ox 25-29 Cu-O chains 445,460, 473,475,482, 483 Cul coordination in 123-O~ 9 Cu2 coordination in 123-O~ 9 CuO3/gaO surface termination 403 CuO 2 bilayers 401 CuO chain 584 CuO 3 chains 401,409 CuO2 planes 381,458, 473,483 Cu spin ordeling 332 Cu spin structures 321 cyclotron frequency 494 -

-

-

-

-

d~2 _ ),2 symmetry 575 d~ v2-wave 601 DO synthesis of 123-O, 13 lattice parameters of samples 44 de Germes factor 362-364 de Gennes scaling 362 Debye-Waller factors of 123-O~ 66 anomalies 65-67 density fluctuations 520 density of charge carriers 365, 368 density of states 569 diamagnetic transition of overdoped 123-O~ I31-133 splitting of 131 diffusive tunneling 575 dimpling (Cu2-O2,O3) transition in 123-Ox 118-127 NPD investigations 118 in overdoped 123-0~ 127 130 displacive martensitic transformation 127 - of Cu2 planes 11 - of planes in 123-Ox 10, 28 -

-

SUBJECT INDEX

-

dipolar interaction 316, 331, 362, 364-368, 376, 381,385 dirty limit 265 displacive martensitic transformation ha overdoped t23-O 4 1t7, 118 distortion of the planes in 123-O~ 68 71 double exchange interaction 368, 380 Drude conductivity 449, 452 Drude feature 452 Drude formula 461,491 Drude-like feature 459 Drude model 452, 454 extended 522 Drude peak 481 Drude response 521 d-wave supercondnctors 476, 478, 488, 489, 491, 574, 586 d-wave symmetry 427 DyBa2Cu30 x 209, 356, 357, 361,364, 373,374, 377-379 DyBa2Cu306 335 DyBa2Cu307 ~ 359, 366 DyBa2Cu307 335, 336, 358 DyBaaCu408 338, 364, 381-383 Dy~Ba4Cu7014 +~ 364, 384 Dy2Ba4CUTOI5 339, 384 Dy2Ba4CusOI6 337, 338 dynamic susceptibility tensor 513, 516 dynamical structure factor 518, 520 ESR (electron spin resonance) 365 EXAFS (extended X-ray absorption fine structure) and PDF (pair distribution ftmction) investigations of the apical oxygen in 123-Ox 7~78 EXAFS investigations of Ca-t23-Ox 159-162 "edge" jtmction 598 effective electron number N~r(e)) 446, 449 effective mass 465 electron attenuation length 403 electron correlations 398 electron-doped superconductor 317 electron energy loss 223 electron mean free path 193 electron-pair botmd states 542 electron radiation 221, 231 electron stopping power 223 electron-boson interaction 455, 458 electron-electron correlations - cuprates 521

-

-

-

679

hexaborides 521 manganites 521 titanates 521 electron-electron interactions 519 electron impurity scattering 519 electron-phonon coupling 451 electron-phonon scattering 519 electronic cluster formation 386 electronic disorder 381 electronic phase diagram 570, 571 electronic sublattice 377 ellipsometry 441 energy gap 481 entropy change 353,381 equilibrium samples of 123-O~ 38 ErBa2Cu30 ~ 374, 377 379 ErBa2Cu307 ~ 356, 357, 359, 361,364, 366 ErBa2Cu307 333 335, 344, 345, 358 gr2Ba4CuTOl4+x 333, 364, 375, 383-385 gr2Ba4Cu7OE4.3 386 Er2Ba4CusO~6 333 etching 585 EuB 6 522, 535 EuBa2Cu307 o 355-357 EuBa2Cu307 354, 528 (Eu I yGdy)Ba2eu307 356 exchange anisotropy 376 exchange constants 379, 380 exchange interaction 364-367, 375, 376, 381, 385 excitonic bound states 542 exotic pairing 318 extended van Hove-like singularity 408 extended-s symmetry 575 4f-hybridization 317, 341 Fano coupling of the superconducting carriers in 123-0~ 63 Fano effects 550 Fano lineshapes 532 Fe doping 357 Fermi liquid 454, 458, 473, 512, 535 Fermi surface 417 extraction from ARPES data 418 - methods 419 - of Y123 418, 421 Fermi-surface harmonics 519, 549 ferromagnetic neutron peaks 368 Ilelll mapping zJ~, zJ~ final-state effects 396 -

680

SUBJECT INDEX

final-state interactions 542 fishtail 188, 204, 207-212, 218fluctuation-dissipation theorem 518 flux line 187 195, 206-209, 212, 214, 218-220, 222, 224, 226-228, 234-236, 241 flux pinning 187-242 - s e e a l s o fishtail - cascade 221-224, 226, 231 - columnar tracks 213, 223-225, 227, 234, 235 defect cascades 213 - fission tracks 213, 225-227, 232, 240 - heavy ions 213,223,234 - neutaon 213, 221-223,225-227, 229 232, 235, 240 precipitate 206, 212-214, 216, 242 freezing of oxygen in 123-O~ structure 84 frequency-dependent scattering rate 453 f-sum rule 520 fully frustrated spin system 339

HgBa2CaCn206 544, 551 HgBa_TCaCu206 + ~ 272, 274 high-resolution electron microscopy in 123-O x 69 2H-NbS 2 superconductor 536 2H-NbSe 2 superconductor 536, 542, 594 HoBa2CU3OT-,5 361,364 HoBa2Cu307 335, 360 Ho2Ba4CusO12 339 hole density 570 hole pockets 422, 423 holon 571 Hubbard model 444, 447 hybridization 340, 360 s e e a l s o 4f-hybridization hydrostatic pressure 355 hyperfine field 354 hyperfine interactions 324, 367

Ga doping 343, 357 gap anisotropy 547 GdBa2Cu30 x 229, 230, 356, 357, 361,364, 372, 373 GdBa2Cu306 +~ 335 GdBa2Cu30614 335 GdBa2Cu307 ,~ 358, 366 GdBa2Cu307 355, 359 Gd2CuO 4 529, 530 generalized susceptibilities 284, 442 Ginzbnrg-Landau fluctuation theory 255, 258 ground-state J-multiplet 353

impurity effects in cuprates Co 593, 594 Ni 593, 594 Zn 593, 594 incommensmate response 297, 307 indirect exchange 368 induced 3D order 336 inelastic scattering 396 infrared conductivity 547 infrared spectroscopy 437-498 c-axis plasmon 479 delta peak 459 irthomogeneities 360, 377 inner vortex excitation 594 - Bi2212 597 botmd states 594 Y123 595 597 in-plane anisotropy 375 in-plane antiferromagnetic correlation length 511 in-plane coherence length 259 insulator-to-metal transition 570 interlayer coupling 337 interlayer hopping 538 interlayer-tunneling models 485 intermediate oxygen range 375 interpiane conductivity 462 intrinsic Josephson junctions 568, 569 - lcR ~ 569

Hall effect 263-276 - angle 266, 267, 274 anomaly 265, 266 - negative 263, 267, 268 carrier number 273, 274 - coefficient 267, 271,272, 274 resistivity 268 negative 264, 267, 271 voltage 263,265, 267 sign change 263 4Hb-TaS~ superconductor 536 heat capacity, s e e specific heat Heisenberg antiferromagnet 522 Heisenberg model 354, 375 Heisenberg spin Hamiltonian 365 HgBa2Ca2Cu308 544, 549, 551 -

-

-

-

-

SUBJECT INDEX irreversibility line 188, 190, 195, 202, 204, 212, 213, 220-222, 224, 225, 227, 232, 234-236, 252, 254, 258-260, 262, 276 Ising model 354, 375 Ising spin dynamics 345 Ising spins 380

-

Josephson effect 485 critical current 599, 601 plasma resonance 439

K2CoF4 336 KK relations, s e e Kramers Kronig relations K2MnF4 529 K2NiF4 336, 522-524, 529 kinetic energy 485 Kramers-Kronig (KK) relations 441-443, 493 Kramers ion 353, 358-360 Nd, Sin, Dy, Er, Yb 358 360 LaBa2Cu30 ~ 354 La2CuO4 287, 290, 321,445, 446, 448, 450, 451, 466, 467, 477, 523, 588, 598 La2CuO4 .,~ 449, 478 La2NiO 4 451 La 2 _ ,(Sr, Ba).,.euO4 285, 292 L%sSr0sCoO 3 535 Lal.~sSrll.lsCu I _xA~O4 274 LaLssSro.lsCuO4 466, 475, 478, 496 La 2 _.~SrxCuO 4 268, 297, 304, 439, 445, 446, 448, 469, 532, 535, 538, 543 545, 549, 550 La 2 ~Sr~NiQ 447 La 2 ~StxNiO 4 +,~ 450 lanthanide magnetic structures 326 330 lanthanide ordered moments 326 330 lanthanide ordering temperatures 326, 327 lattice distortions in 123-O~ 67, 68 lattice effects in 123-O~ 44, 64-78 lattice parameter changes with x 22, 39, 44 split atom model of the chains 65 lattice parameters o f / 2 3 - O , BAO samples 44-49, 52 - CAR samples 39-43, 50, 51 change with x 22-49 - equilibrium samples 38-49 - non-equilibrium samples 22-25 DO ~am~les 44--49. 53 Lindemann criterion 262 Ln 2 ~Ce, CuO4_y 255 -

-

-

-

-

681

local oxygen ordering in 123-O., after quenching 35 local structm'e of 123-O~. 64-78, 12(~127 EXAFS investigations 120 EXAFS investigations of the Cu2 planes 120 localized carriers in Ca-123-O,. 163 long-range magnetic order 371,375 longitudinal screening 546 LuBa2Cu307_ (~ 355, 357 LuBa2Cu307 355 -

-

MFL, s e e Marginal Fermi Liquid magnetic characterization of Ca-123-O~ 156-159 magnetic clusters 375 magnetic correlation length 380 magnetic exchange constants 361,376 magnetic field dependence of specific heat 382 magnetic form factor 319 magnetic impurities 355 magnetic moments 364 magnetic ordering 352, 358, 360, 381 models 360-367 pressure dependence 359 magnetic ordering temperatures 363, 365 magnetic sta'ucture factor 319 magnetic structures 326 - for RBa2Cu3Q+ ~ 335, 337 - for RBa2Cu40 ~ 338 magnctoresistance 251-276 magnon-magnon interaction effects 523 manganites 521,535 Marginal Fermi Liquid (MFL) 467, 522, 534, 535 martensitic transformation 127-130 - in overdoped 123-OX 127 - ofFe I ~.Pd~. 59 mean-field theoretical model 255 mean-field upper critical field 255 mean flee path 396 mean-squared thermal vortex fluctuation displacement 262 melt-textured 188, 196, 200, 201,203, 204, 212, 213, 216, 218-220, 226, 227, 231,232, 238, 242 melting line 262 metal insulator transition 375, 380, 394 metallic re~ion 571 methods ot synmesls oF IZ.5-t)~ BAO 14, 15

682

SUBJECT INDEX

methods of synthesis of 123-O4 (cont'd) - CAR 15 DO 15 4 1 m e V peak 472 microwave spectroscopy 470, 471 miscibility gaps in 123-O~, possibility of 85 mixed state 187-189, 191,228, 253, 263, 265-268 mixed termination 403, 406 MnF 2 522 mobile charge carriers 368 models for magnetic ordering 360-367 moderately clean superconductors 265 molecular field 354 momentum space density of states 420 Monte-Carlo evaluation 362 M6ssbauer spectroscopy 364 Mott-Hubbard transition 395 Mott insulator 331 multicomponent model 452 multilayer effects 550, 551 multilayered systems 330 NMR (nuclear magnetic resonance) 365, 467, 469, 571 Nb 266 Nb 3Sn 542-544 NdBaCu 3 yGayO~ 371 NdBa2Cu30 x 204, 212, 213, 218, 232, 356, 357, 361,364, 369-371,376 NdBaaCu306 ~ x 335 NdBa2Cu3 O6.90 384 NdBa2Cu3OT- 0 359 NdBa2Cu30 v 358 Nd I +yBa~_),Cu30~ 356, 369, 371 Nd~.85Ce0.~sCuO4 j, 264, 266 Ndl.85Ce015CuO 4 268 Nd 2 yCeyCuO4 445, 446, 475, 476, 484 Nd2 ~Ce~CuO4 329, 527, 551 Nd2 yCeyO4 ~ 450 Nd2CuO4 ~ 449 Nd2CuO 4 .y 449 Nd2CuO 4 3 2 1 ~ 2 4 , 344, 451,523 Nd noncollinear spin structures 324 N6el temperature 332 neutron diffraction 189, 191,352, 358, 360, 362, 367, 368, 375, 38i neutron irradiation 212, 221,222 neutron scattering magnetic ordering 315-346

spin fluctuations 2 8 1 ~ 1 2 neutron spectroscopy 358 Ni-doped La 2 ~Sr~CuO~ 273 non-Fermi liquid 345 non-Kramers ions (Pr, Ho, Tm) 360 noncollinear spin structures 318, 321,322 nonequilibrium synthesis of 123-Ox 12 nonsuperconducting PrBa2CusO 7 360 normal-state gap 468 normal-state pairing 541 normal-state pseudogap 439 nuclear magnetism 355 nuclear spin ordering 324, 328 -

one-component model 452, 456 1-2-3 systems 331 Onsager 335 onset of superconductivity in 123-O4 42 optical conductivity 521 optimally doped cuprates 457, 471 - synthesis of 15, 16 order parameter 547, 566, 575 - admixture of s-wave 601 s-wave 594 - symmetry 586, 587, 594 temperature dependence 587 ordered magnetic moment 319 organic superconductors 257 ortho-Ii phase 402 orthorhombicity 29, 33-36, 39, 40, 46-48, 50 53, 109, I l l , 125, 126, 174, 175, 361 ortho-V superstructure in 123-Ox 82 ortho-Vlll superstructure in 123-0~ 82 oscillatory magnetic states 316 overdoped cuprates 145 165, 457 Ca doping 145 oxygen and Cu orbitals 360 oxygen concentration 331 - electron doped 328 oxygen content of Ca-123-O~ 146 ageing effects 161, 162 change with effective Ca charge 162 oxygen content of 123-O~ determination 17-20 high-accuracy volumetric method 18 iodometry 17 - thermo~ravimetric reduction 17 thermogravimetry 17 oxygen deficient 208-212, 219, 228, 242 -

-

-

-

-

-

SUBJECT INDEX oxygen dependence 339 oxygen diffusion in 123-O,. 135-138 oxygen mobility in 123-Oz 33-38 oxygen nonstoichiometry in YBa2Cu3Ox 1-176 - change of bond length with x 22-49 change of lattice parameters with x 22-49 oxygen relaxation and T~ increase after quenching 35 oxygen reordering at quenching in 123-O, 33-38

-

PDF (pair distribution function) investigations in 123-O5 67 P - T cross section of pseudo-binary P - T - x phase diagram of 123-Ox 103 pair breaking 546, 603 pair formation 512 pairing mechanism 392, 587 pancake 191, 195, 208, 209, 212, 224, 228, 236, 241 paramagnetic critical scattering 290 Pb2SrzPrCu308 344 Pb2Sr2RI _xCa~Cu308 340 Pb2SrzReCa_~Cn3Q 478, 492 penetration depth 441,477, 481-483, 485, 486, 488, 492 phase coherence 512, 573 phase diagram (hard X-rays) of 123-O~ 79-87 model calculations 87-93 phase diagrams 376, 377, 511 - 123-Ox significant points of the T - x section 165-173 phase fluctuations 547 phase-sensitive measurements parallel tunneling 598-600 tunneling 600, 601 phase separation in 123-O~ 131 145 Ca-overdoped phase 145 - overdoped phase 131 Raman investigations 110-117 Raman phonon splitting 154 splitting of the diamagnetic transition 138 stripe formation 141 phase separation in YBa2Cu30 ~ 1-176 phase transitions in Y-BazCu30~ 1-176 phenomenological Lindemann melting criterion 262 phonon-assisted hopping 538 -

-

-

-

-

p

e

r

p

e

n

d

i

c

u

I

a

r

683

phonon modes 462 photoelectron diffraction 419 photoemission spectroscopy 397 photon energy dependence 408 (~,zc) resonance 309-311 pinning centers 213-227 - s e e a l s o flux pinning dislocation 209, 213,216~20 "R-phase" model 323 planar-type jmaction 567 plane copper 401 plane oxygen 401 plasma frequency 520, 521, 547 plasma resonance 496 PmBa~Cu30x 354 point contact 567 polarization of Nd 325 polarization of Pr 325 polaron 450 PrBa2Cu2NbOs 343 PrBa~Cu30z 356, 357, 360, 364, 368, 369 superconductivity 341 PrBa2Cu306 +x 335, 340 - hysteresis 342 PrBa2Cu307_,~ 361 PrBa2Cu307 343,360, 526 PrBa~ Cu40 ~ 343 Pr2_yC%CuO4 446 Prl.sCe0.5Sr2Cu2(Nb,Ta)Oio 344 Pr2CuO4 323,344, 451,529 Pr 4d resonance 400 pre-formed pairs 587 pressure dependence - entropy 383 magnetic specific heat 382 - of T~, in 123-Ox hy&'ostatic 99 - uniaxial 105 pressure measurements 356, 360, 381 pseudogap 464-466, 468, 469, 478, 493, 511, 538-541,566, 570-574 - anisotropy 574, 575 temperature dependence 590-592 -

-

-

-

-

-

-

-

quantum fluctuations 262, 263,329 quantum-mechanical Monte-Carlo routine 367 quasi-two-dimensional (2D) vortex fluctuations 260 quasiparticle dispersions 414 quenching of 123-O, 12

684 quenching of 123-O, ( c o n t ' d ) increase of T~ after 33 relaxation of oxygen after 35 quenching of the total angular momentum

SUBJECT INDEX

353

RBa2 Cu30 ~ 352 Dy 375 - Er 375 - Gd 375 Nd 375 Pr 375 - Sm 375 RBa2Cu306 532 RBa2Cu3Q+x 317, 331 RBa2Cn307_,~ 265, 402 RBa2Ctl307 333,360 RBa2Cu408 381 R2Ba4CuTOi4 +~ 383 R2BaaCUTOi5 331,337 R2Ba4CusOI6 317, 331, 337 R2CnO 4 318, 319, 531 - b c t lattice 319 b c t symmetry 329 RKKY (Ruderman-Kittel-Kasuya-Yosida) interaction 362, 364, 365,368, 372, 387 RM%S8 316 RNi2BaC 317 Rl_~Pr~.Ba2Ctt307 ~ 256 RRh 4B4 316 RVB (resonating valence bond) 571 radiation-induced defects 221,224 Raman-active excitation 516 Raman apex phonon frequency vs. x, in 123-O~ 63 Raman apex phonon width, deconvolntion in 123-O~ 113 Raman Big phonon of 123-O~, change with x 63 Rarnan characterization of Ca-123-O,. 154-156 Raman m-phase phonon of 123-O., as a function of x 63 Raman investigations in 123-O~ 62-64, 110-117 apex Raman phonon and phase separation I10 near the T O Iransition of 123-O~ 62 softening of the in-phase 02, 03 phonon 119, 120 Raman scattering 512 522 - c-axis electronic scattering 536

charge and spin ordering 539 chiral spin excitations 513 chiral spin fluctuations 529 -- collision-dominated scattering 521,522, 534 - cross-section 514, 515, 520, 544, 549 transverse screening 546 crystal-field excitations 513, 532 effects of optical response 515 electronic Rarnan scattering 518, 521 excitonic scattering 530 excitons 513 high magnetic ficlds 553 magnon scattering 515 "normal" metallic phase 532 phonon 515 qnasielastic Raman scattering 530 Raman response function 518, 544 Raman scattering vertex 518, 519, 521, 527, 535, 540, 547, 548 resonance effects 516, 524 - resonant X-ray Raman scattering 531 scattering Hamiltonian 514, 522, 529 selection rules 516 spin-phonon interactions 524, 528, 531,532 superconducting gap excitations 542, 544 superconducting state response 548 two-magamn 5ll, 513 two-magnon excitations 514 two-magnon Raman scattering 539 two-magnon scattering 522 two-phonon excitations 514 Raman spectroscopy 62 64, 468, 469, 509-555 Random Phase Approximation 546 reentrant superconductivity 316 relaxation 195-197, 199, 201,208, 209, 211, 228, 239 relaxation of oxygen after quenching 35 residual absorption 475, 476 resistivity 571 resonance photoemissinn 399 resonant X-ray Raman spectroscopy 555 reversible magnetization 254, 259 rods of magnetic scattering 333,334, 336 -

-

-

-

-

-

-

SIS (superconductor-insulato~superconductor) junction 586 sample inhomogeneity 358 sawtooth profile 338 scanning tunneling m~croscope t~ xivi) )o~-ou,~ scanning tunneling spectroscopy (STS) 575

SUBJECT INDEX scattering rate 455-457, 465, 467-469, 471, 475 c-axis 464 Schottky anomalies 353 semiconducting samples 375 shadow bands 418 - inY123 424 short-range magnetic interactions 373, 375 short-range order 381 single-component model 453 single crystals of cuprates I88, 191, 196 198, 204, 207-212, 216, 218, 219, 221,222, 227, 228, 231,234, 235,240, 242, 357 single-layer systems 318 size of the gap 587 slightly doped systems 380 SmBa2Cu;O, 356, 357, 361,364, 371,372, 378, 379 SmBa2Cu3069 o 384 SmBa2Cu3Ov ,7 359 SmBa2Cu307 358, 375 SmlssCeolsCuO4_., 255,257, 260, 263 Sm2CuO4 320-322, 523 small domains 328 specific heat 351-387, 572 - electronic 352 - hyperfine 354 - lattice 352 - magnetic 353, 358, 359, 361,370-374, 378, 384-386 - - field dependence 382 pressure dependence 382 measurements 365 - ofRBa2Cu307 354-360 - - Ce 354 - - Eu 354 - - Gd 354 - - La 354 - Lu 354 - - oxygen-deficient RBa2Cu30 , 367-375 - - Pm 354 - - Tb 354 - - Y 354 spectral function 420, 454, 455 spectral weight 446, 447, 471,475, 478, 481, 486 spectroscopies - ARPES 391-432, 467, 469, 471, 574 - infrared 437-498 - Raman 509-555

-

-

-

-

-

-

685

tunneling 563-604 specular tunneling 575 spm-½ systems 354 spm cluster 366 spin depairing 316, 326 spin direction 319 spin dynamics 344 spin excitation 572 spin-flop transition 342 spin fluctuations 282 spin gap 571 573 spin-reorientation transition 324 spin singlet 587 spin waves 286 dispersion 345 spin-charge separation 472, 587 spin-orbit interaction 329 spinon 571 Sr2CuO 3 407 SlhCuO~C12 407, 451,523, 524, 526, 531 Sr doping 356 Sr,.La~ _xTiO3 522, 535 Stokes scattering 513 stripe 472 stripe formation in 123-O., 141-145 striped phase 380, 472 structural characterization of Ca-123-Ox 151-154 structural distortions in YBa2Cu30 x 1-176 structural evolution of 123-Ox with nonstoichiometry 11 64 structural instability of 123-O~ 32 structural optical and magnetic anomalies at x,~6.75 of 123-O~ 108-110 structural parameters of BAO samples of 123-O~ 54 - of highly overdoped YBa2Cu30 x 10 structural T ---*O transformation 22-49 at insulator-superconductor transition of 123-O., 58-62 structm'e of 123-O~., average crystallographic 8-11 sublattice magnetization 333 sum rule 443,481 superconducting critical temperature 255 superconducting delta function 478, 481 superconducting gap 426, 441,468, 473, 474, 541,542, 565 carrier concentration dependence 588 gap structure 584-588 -

-

686

SUBJECT INDEX

superconducting gap (cont'd) impurity effects 593, 594 in-plane anisotropy 597 - strong coupling 565 temperature dependence 590-592 - weak coupling 565 superconducting transition temperature, see T~ of 123-O~ superconductor-insulator-normal metal (SIN) function 575 superexchange 329, 362, 365, 368, 387 and RKKY interaction 364 coupling constant 511 superfluid density 490 superstructures in 123-Ox 79 2a0, 3a 0 random faulting sequence 86 2% (ortho-II) 81 3a 0 (ortho-III) 81 calculation with structure combination branching 90 ortho-V 82 ortho-VIII 82 possibility at steps of in-phase phonon 116 surface chain feature 409 surface resonance 409 surface sensitivity 397 surface termination 431 susceptibility tensor 515 s-wave superconductors 476, 491,586 s-wave symmetry 575 synthesis conditions - for equilibrium samples of 123-0 x 13 for quenched samples of 123-0x 27 synthesis near thermodynamic equilibrium of 123-0~ 13, 14 synthesis of overdoped samples of 123-0 s 14, 15 synthesis of underdoped samples of 123-0 x 15, 16 -

-

-

-

T O transition of 123-Ox 51-64 Raman investigations 62 T~ of 123-O~ 93-106 - as a function of x 5 dependence on hole concentration 93 - dependence on nonstoichiometry 93 generic parabola 96 doubts about validity m Ca-123-Ox 164 - giant dTJdP effects at x ,-~ 6.75 and x ~ 6.4 99 -

-

increase after quenching 33 - pressure dependence 99 - hy&ostatic 99 uniaxial 105 splitting with Ca doping 157 T ~ phase 318 T' structure 319 2D behavior 317-319, 332, 344 2D Ising antiferromagnet 333, 339 2D Ising model 335, 336, 354, 361,381 2D magnetism 332, 337 2D magnets 317, 318 t - J model 571 TbBa2Cu3Ox 354 TbSr2Cu2.69Mo0.307 335 terahertz spectroscopy 493 terahertz time-domain spectroscopy 442 thermal fluctuations 262, 263, 272 thermodynamic fluctuations 255 thin films 188, 191, 196, 198, 209, 213, 216-218, 224, 227, 228, 231, 232, 235, 236, 242 three-dimensional (3D) ordering 367 three-dimensional (3D) vortex fluctuations 260 three-step model 396, 397 T12Ba2Ca2Cu3Olo 267, 271,272, 532, 543 T12Ba2Ca2Cu3Oi0 + ~ 569 T12Ba2Cao 9~Gd0o2Cu20 ~ 450 T1BazCaCu207 536 T12Ba2CaCu208 264-266, 269, 271,536 TlzBa2CuO 6 255, 256, 260, 492, 532, 544, 545, 549-554 TlzBa2CuO6 + 6 439, 451,458, 475, 478 T1BazPrCu207 343 TI cuprates 274 TmBa2Cu307 ,~ 356, 357, 361 TmBa2Cu307 360 transition from metallic to semiconducting behavior 368 T* 564, 570 tunneling process 575, 576 tunneling spectroscopy 469, 483,563-604 tunneling spectrum 578 - background 579, 582 tweed microstrucmre at a martensitic transformation 58 twin boundary 601 twin planes 213,216, 217, 219 2-4-7 systems 331 2-4-8 systems 331 two-magnon excitations 527 -

-

-

-

SUBJECT INDEX tmderdoped cuprates 457, 465, 469, 471 untwirmed crystals 401 upper critical field 252, 255, 256, 258, 276 V 266 V3Si 542, 544 valence band YBCO 404 very clean limit 265 vortex core 572 vortex dynamics 495 vortex viscosity 495 W e r t h a m e r - H e l f a n d - H o h e n b e r g

M a k i

( W H H M )

theory 253,259 work function 396 X Y model

373,374

Y123/Au/Pb junctions 599 (YBazCu307 0)J(PrB~Cll307 ,~),,, superlattices 264, 271 YBa2(Cul ,Co,)307 o 270 YBa2(Cuj ~Ni~)307 ~ 270 YBa2Cu3 .,Co~O7- o 274 YBa2Cu3 xFexO7- 6 274 YBa2Cu30~ 1-176, 188-191, 203, 208, 209, 211-213,216-221,223, 224, 226-228, 231,232, 234, 236, 240, 259, 260, 263~66, 268-270, 274, 354, 439, 445, 448, 451,452, 456, 459, 460, 462~64, 466, 467, 472, 475, 477, 478, 482, 485, 488, 489, 492, 493, 566, 582 584 (001)-oriented plane 583, 584 (ll0)-oriented plane 584 (100) surfaces 598 (110) surfaces 598 anomalous transitions at x ~ 6.75 108 - thin films 440 tunneling spectrum 587, 588 YBa2Cu3Oc, 402, 451,523, 524, 528

YBa2Cu~O6 + o 524 YBa2Cu~O6 +.~ 294, 449, 527, 534, 536-54I YBa2Cu306.1 525, 526 YBaaCu3O6.3 450 YBa2Cu~O6.6 493 YBa2Cu307 x 305 YBazCu306.95 471,474, 478, 481,485, 486, 489, 492 YBa2Cu307 o 253, 263, 264, 268, 271-273, 355, 400 YBa2Cu307 265,465, 532, 535, 536, 543, 544, 549 average structure of orthorhombic 9 YBa2Cu408 412, 445, 451,459, 460, 462, 475, 477, 478, 482, 485,492, 527, 544 YBa2Cu307_ JPrBa2Cu307 ~ multilayers 271 YBa2Cu307/PrBa~Cu307 multilayers 264 YBa2Cu3 .,Zn~O7 ~ 256, 273, 274 Yl -xEr~Ba2Cu307 357 Y~ _ xPr~Ba2C% _yA1),O7 527 yu 5oPro.soBa2Cu3O6.95 405 Yos~Pro.42BazCu307 6 268 Y1 - ~PrxBa2Cu306.95 400 Y~ _xPl~Ba2Cu306.97 259, 262, 263 YI-~Pr~Ba2Cu307 o 255, 264-266, 269, 270, 273~75, 402, 412 Yi-~Pr, Ba2Cu307 356, 527, 539 Yo.92Proo~Ba2Cu307 o 268 YbBa2Cu307_ ~ 197, 253, 356, 357, 359, 361, 364 YbBazCu307 335, 358 -

-

-

-

-

-

-

687

zero-bias anomaiy 565, 597 zero-bias conductance peak 586 splitting 601-603 zero-energy states 598 Zn doping 342, 357 Zr-gettered samples of 123-O, 27, 51 structural parameters of 24

-

-

Handbook on the Physics and Chemistry of Rare Earths. vol.31 High-Temperature Superconductors - II

Read more

Handbook on the Physics and Chemistry of Rare Earths. vol.30 High Temperature Rare Earths Superconductors I

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

Handbook on the Physics and Chemistry of Rare Earths

Read more

High Temperature Superconductors

Read more

High-Temperature Superconductors

Read more

Handbook on the Physics and Chemistry of Rare Earths. vol.10 High Energy Spectroscopy

Read more

Handbook on the Physics and Chemistry of Rare Earths, Volume 37: Optical Spectroscopy (Handbook on the Physics and Chemistry of Rare Earths)

Read more

High Temperature Superconductors

Read more

High Temperature Superconductors

Read more

Handbook on the Physics and Chemistry of Rare Earths. vol.29 The Role of Rare Earths in Catalysis

Read more

Handbook on the Physics and Chemistry of Rare Earths. vol.18 Lanthanides-Actinides: Chemistry

Read more

Handbook on the Physics and Chemistry of Rare Earths. vol.17 Lanthanides-Actinides: Physics I

Read more

Bismuth-based High-temperature Superconductors

Read more

Bismuth-based high-temperature superconductors

Read more

Microstructure and Properties of High-Temperature SuperConductors

Read more

Vortices in high-temperature superconductors

Read more

Symmetry and Heterogeneity in High Temperature Superconductors (NATO Science Series II: Mathematics, Physics and Chemistry)

Read more

Handbook on the Physics and Chemistry of Rare Earths, Volume Volume 20

Read more

Handbook on the Physics and Chemistry of Rare Earths, Vol. 33

Read more

Handbook on the Physics and Chemistry of Rare Earths, Volume Volume 22

Read more

Handbook on the Physics and Chemistry of Rare Earths. vol.35

Read more

Recommend Documents

Handbook on the Physics and Chemistry of Rare Earths. vol.31 High-Temperature Superconductors - II

Handbook on the Physics and Chemistry of Rare Earths. vol.30 High Temperature Rare Earths Superconductors I

Handbook on the Physics and Chemistry of Rare Earths

Handbook on the Physics and Chemistry of Rare Earths, volume 13 Elsevier, 1990 Edited by: Karl A. Gschneidner, Jr. and L...

Handbook on the Physics and Chemistry of Rare Earths

Handbook on the Physics and Chemistry of Rare Earths volume 8 Elsevier, 1986 Edited by: Karl A. Gschneidner, Jr. and LeR...

Handbook on the Physics and Chemistry of Rare Earths

HANDBOOK ON THE PHYSICS AND CHEMISTRY OF RARE EARTHS Advisory Editorial Board G.-y. ADACHI, Kobe, Japan W.J. EVANS, Ir...

Handbook on the Physics and Chemistry of Rare Earths

Handbook on the Physics and Chemistry of Rare Earths, volume 14 Elsevier, 1991 Edited by: Karl A. Gschneidner, Jr. and L...

Handbook on the Physics and Chemistry of Rare Earths

Handbook on the Physics and Chemistry of Rare Earths, volume 15 Elsevier, 1991 Edited by: Karl A. Gschneidner, Jr. and L...

Handbook on the Physics and Chemistry of Rare Earths

HANDBOOK ON THE PHYSICS AND CHEMISTRY OF RARE EARTHS Advisory Editorial Board G.-y. ADACHI, Kobe, Japan W.J. EVANS, Irv...

Handbook on the Physics and Chemistry of Rare Earths

Handbook on the Physics and Chemistry of Rare Earths, volume 32 Elsevier, 2001 Edited by: Karl A. Gschneidner, Jr., LeRo...

Handbook on the Physics and Chemistry of Rare Earths

Handbook on the Physics and Chemistry of Rare Earths, volume 33 Elsevier, 2003 Edited by: K. A. Gschneidner, Jr., J.-C. ...